Professional Documents
Culture Documents
Contributions 16 Ghent Altarpiece Interior Lower Register 2021 Searchable Mid
Contributions 16 Ghent Altarpiece Interior Lower Register 2021 Searchable Mid
The Ghent Altarpiece by Hubert and Jan van Eyck has always attracted both admiration
THE GHENT ALTARPIECE – Research and Conservation of the Interior: the Lower Register
and curiosity. Following the treatment of the closed polyptych, we turned to the paintings
of the lower register of the open altarpiece, which includes the famous Adoration of the
Lamb. A major research and restoration campaign by the Royal Institute for Cultural
Heritage (kik-irpa, Brussels) was carried out between 2016 and 2020.
The findings reflect those from the treatment of the closed altarpiece: large areas of the
panels were covered with overpaints, the vast majority dating from the 16th century. Their
removal allows the Van Eycks brothers’ paintings to be truly appreciated again.
The three years that the conservators spent working on the panels enabled them to
examine the paintings down to the smallest detail. Interdisciplinary collaboration with
laboratory scientists, art historians and experts in scientific imagery led to a series of
discoveries – some anticipated in light of earlier research, others completely unexpected.
The authenticity of the quatrain stating that Hubert had begun the work, that his
brother Jan completed it and that it was presented in 1432, has been firmly established
recently. New hypotheses are proposed here regarding the division of labour between the
two Van Eyck brothers. The contribution of Hubert, who died in 1426, can finally be seen
more clearly and turns out to be far from negligible. But the removal of overpaints has
revealed non-Eyckian interventions too...
ISBN 978-2-39025-441-4
THE GHENT ALTARPIECE
Research and Conservation
of the Interior: the Lower Register
The Ghent Altarpiece
Research and Conservation of the Interior:
The kik-irpa produces three series of scientifi c publications on 15th- and early
1 6th-century paintings: the Corpus, the Repertory and the Contributions.
d/20021/0
0613/1
isbn 978-2-93000054-41-4
Main Authors
Griet Steyaert, Marie Postec, Jana Sanyova, Hélène Dubois
Painting Conservators-Restorers
Kathleen Froyen, Nathalie Laquière, Laure Mortiaux, Françoise Rosier,
Bart Devolder, Cécile de Boulard
Foreword
Hilde De Clercq vii
Introduction 1
2. The Van Eycks’ Creative Process and the Different Stages in the Execution
of the Interior Lower Register of the G HENT ALTARPIECE
Marie Postec, Griet Steyaert
With contributions by our colleagues, the researchers and specialists mentioned on the title page 33
Conclusion 1 73
Bibliography 1 93
Abbreviations 2005
V
Foreword
When the conservation project of the Van Eyck brothers’ Adoration of the Mystic Lamb
– the Ghent Altarpiece – started in 20012, expectations ran high. They were certainly
met, and even exceeded, thanks to the extraordinary results of the treatment and
of the associated scientifi c research. In the course of the fi rst phase, the project was
signifi cantly modifi ed due to the unexpected discovery – during varnish removal
on the exterior of the polyptych – of overpaints covering the major part of the
surface. Thanks to the fl exibility and the confi dence of the Church Council of
St Bavo’s Cathedral and generous funding by the Flemish Government and the
Baillet Latour Fund, the same rigorous approach could be continued during the
second phase of the project. Work was carried out in public in the purpose-built
glazed workshop at the Ghent Museum of Fine Arts (msk), which also generously
supported the project and the team on a daily basis.
As in any other conservation project undertaken by the Royal Institute for Cultural
Heritage (kik-irpa), an interdisciplinary approach was applied, involving highly
experienced conservator-restorers, scientists and art historians, always keeping up
with the latest research in their own fi eld. Results of the conservation and research
carried out during the second phase of the treatment (20016 –19) are published here.
They concern the lower register of the altarpiece interior: the eponymous central
panel of the Adoration of the Lamb, the Knights of Christ, the Hermits and the Pilgrims.
Paul Coremans, the founder of our institute and instigator of the conservation of
the Ghent Altarpiece in 19500 – 51, was a visionary. His dedication to interdisciplinary
research and fascination for the painting technique of the Flemish Primitives led to
pioneering results. Seven decades later, Coremans’s legacy certainly lives on in our
interdisciplinary research and collaboration with Belgian universities and foreign
institutions while new analytical and imaging techniques are being developed, and
samples are even analysed in synchrotron facilities. This intense collaborative effort
lifts the veil on a persistent enigma in the history of Western art: the Van Eyck
brothers’ respective shares in the conception and execution of the Ghent Altarpiece,
and their fascinating technical ‘secrets’.
This book was written in the transitory period between the second and third
phases of the conservation and restoration project. It is based on fi rst-hand
information gathered by the conservators and scientists who have been in direct
contact with the paintings for several years. They have acquired a unique knowledge
of the Van Eycks’ painting technique and their subtle variations of paint handling.
Combined with the laboratory-based scientifi c analyses and the art-historical
VII
FOREWORD
facts, their fi ndings will transform scholarly – and popular – perceptions of the
achievements of Hubert and Jan van Eyck.
I would fi nally like to extend my sincere gratitude for the sustained collaboration
with research partners and institutions, in particular with the universities of
Antwerp and Ghent. We are also extremely grateful to the advising bodies who
are too numerous to be listed here,1 to the Gieskes-Strijbis Fund, which sponsored
research, and to the Périer-D’Ieteren Foundation for its support towards the present
publication.
Hilde De Clercq
Acting Director - General, kik-irpa, Brussels
Notes
1 A list of project participants is (20012–17); http://closertovaneyck.kikirpa.
published on the website ‘Closer to be/ghentaltarpiece/#home/sub=credits
Van Eyck’: http://closertovaneyck. (20018–200); see also Project Participants
kikirpa.be/ghentaltarpiece/#home/ in this volume.
sub=credits&goto=credits20012
VIII
IRR IRR/XR 10 VIS/XR/IRR
VIS/XR/IRR 6
VIS/IRR VIS/IRR
7
11 12
VIS VIS/IRR
1 13
8
VIS/IRR
VIS/IRR
2
VIS 9
3 VIS/IRR
VIS/IRR
VIS/IRR
4
14
5
XR/IRR
VIS/IRR
15
XR VIS/IRR
VIS
16 26
25 VIS
VIS/IRR
VIS/IRR 30
VIS VIS XR
29
17 20 XR
22 VIS/IRR
18 IR
VIS
21 24
23 28
27
VIS/IRR
VIS
19
31
X
Note to the Reader
Abbreviation ‘cve’
is often used to describe a later style derived from (Jan) van Eyck’s work, but also
in connection with Jan van Eyck or the Van Eyck brothers.’2 In this volume, we
use the terms ‘Eyckian’ for everything that could have been executed by Hubert
or by Jan, and ‘non-Eyckian’ for everything that was not executed by either of the
two brothers. However, this does not exclude the intervention of a contemporary
painter.
Cross-section numbering
Notes
1 Deneffe et al. 200
009, pp. 16–17. 2 Smeyers 1996; Verougstraete, Van Schoute
2000
000; Stroo, Martens 200200, p. 353.
XI
To the memory of J.R.J. Van Asperen de Boer
Introduction
Restoration campaigns offer a privileged opportunity to study works of art. The Van
Eyck brothers’ Ghent Altarpiece was one of the fi rst works for which restoration was
followed by a scholarly publication: L’Agneau mystique au laboratoire, edited by Paul
Coremans, appeared in 1953, following the treatment performed in 19500–51 at the
Laboratoire central des musées de Belgique, kik-irpa’s precursor.1 That publication
and A Scientific Re-examination of the Ghent Altarpiece, written by J.R.J. Van Asperen
de Boer in 1979, are still landmark references2 among the technical and material
studies devoted to the polyptych. While the authors of both publications provided
a great deal of highly pertinent information and observations, their interpretation
has often remained incomplete for two main reasons. Firstly, they did not have access
to the research tools we can call on today, while secondly and most importantly,
they lacked one key piece of information that hindered their conclusions: the fact
that the panels were largely covered by very early overpaints,3 mostly dating back
to the sixteenth century.4
The current conservation and restoration project began in 20012, preceded by a
preliminary study in 20010.5 Given the size of the Ghent Altarpiece, the intention
from the outset was to execute the project in three successive phases: a fi rst phase
devoted to the study and treatment of the paintings of the altarpiece exterior;
a second focusing on those of the lower register of the altarpiece interior; and,
lastly, a third phase to treat the paintings of the upper register of the altarpiece
interior. The fi rst phase was carried out between 20012 and 20016 and the second
between 20016 and 20019, both in public at the Museum voor Schone Kunsten (msk)
in Ghent.6 The third phase, which will bring this ambitious programme to a close,
has yet to commence.
Signifi cant resources were deployed from the outset to fund the restoration
process but also the research and sharing of information that occurred in parallel,
beginning with the placement of important documentation online through
the ‘Closer to Van Eyck’ website. Information of major importance was already
presented at the symposium organized in 20012 by kik-irpa and its Centre for the
Study of the Flemish Primitives.7 Indeed, dendrochronology had established that
the Ghent Altarpiece had been conceived from the outset as a polyptych comprising
twelve panels8 and that the lower register had not been executed separately and
then integrated by Jan at a later date in a larger ensemble, as certain authors
had suggested.9 The fi rst phase of treatment also revealed that the panels had
been substantially overpainted. The restoration treatment of the huge ensemble
1
INTRODUCTION
undertaken in 20012 has involved the removal of these overpaints, revealing the Van
Eycks’ painting, which had been hidden for several centuries. All these data opened
up fresh new perspectives in the study of this world-famous polyptych.
A scholarly publication, The Ghent Altarpiece – Research and Conservation of the
Exterior,10 followed the fi rst phase of conservation and restoration. It describes the
progressive treatment that led to the removal of overpaints, as well as the technical
and scientifi c study of the paintings and their frames. The latter came as a genuine
revelation during this round of treatment. Restoration of the frames also allowed
investigation and confi rmation of the originality of the quatrain11 painted on the
frames of the closed altarpiece, which mentions the participation of the two Van
Eyck brothers. Ever since the quatrain was discovered in 1823, one of the key
questions intriguing the experts has been the respective participation of the two
brothers. The quatrain states that Hubert began the work, that his brother Jan
completed it and that it was presented in 1432. We know, moreover, that Hubert
died in 1426.12 Having previously been contested,13 Hubert’s existence and the
authenticity of the quatrain are both now generally accepted. Taken in combination
with the dendrochronological results, this information suggests that prior to his
death Hubert had played an important part in the genesis of the altarpiece and
possibly in its initial execution.
The second restoration phase, conducted between 20016 and 20019, concerned
the lower register of the open altarpiece. It allowed both the research and the
removal of overpaints to be continued. As Coremans and Van Asperen de Boer
had already noted, it quickly became apparent that the paintings in this zone
of the altarpiece showed evidence of successive stages of development as well as
reworking, indicating that numerous changes were made before the paintings were
fi nalized: far more numerous indeed than had been expected. In-depth technical
and scientifi c research continued during this second phase, supported by the
International Advisory Commission,14 which has supervised the treatment since
the beginning of the project. Its views and encouragement have been invaluable:
The team’s new discoveries of a complex history of very early revisions to the altarpiece
have raised intriguing questions: in addition to the work of Jan van Eyck, do we now
see evidence for the work of Hubert van Eyck or the Eyckian workshop? Hints that the
production of the altarpiece may have extended over some time – that some passages may
have been extensively revised during the creative process, others may have been painted
over early damages, and that there may be evidence for more than one artist – will spur
new scholarship for years to come…15
It made sense, therefore, to continue the publishing effort begun in 1953 and
resumed in 200200 following the treatment and research carried out on the paintings
and frames of the altarpiece exterior. The approach presented here deliberately
differs, however, from the book published in 200200, a substantial proportion of
which was devoted to the overpaints and their removal.
This volume places the emphasis on the pictorial layers of the newly restored
paintings, given the importance of the discoveries. The work of removing the
overpaints and its specifi c challenges, which are described in the fi rst chapter of
2
INTRODUCTION
this book, gave us the opportunity to view the work in a state much closer to the
original. Thanks to the removal of thick yellow varnishes and of the sixteenth-
century overpaints, we can now make out more clearly the details of execution
and the superimposition of paint layers and better appreciate the differences in
quality. The study of these layers and the different stages of execution that have
been observed is detailed in the second chapter, which is devoted to the technical
execution of the paintings. The laboratory’s contribution (chapter 3) focuses on the
analysis of the paint layers, together with the results of the study of the materials
used in the paintings of the altarpiece exterior.16 Certain results regarding the
upper register are also touched on, based on the samples retained from the 19500–51
restoration campaign. The book concludes with a stylistic study (chapter 4), which
proposes attributions based on the new technical and scientifi c data revealed here.
Research into the supports, frames and inscriptions is not included at this point:
the results will be presented in the publication that will round off this immense
project following completion of the third phase. Detailed conclusions in this regard
will only be possible after the restoration of the upper register.
All the new data gathered during and after treatment has been made possible by
a great deal of teamwork. The time and the number of experts assigned to the
project have been important factors in allowing treatment and in-depth study on
such a large scale – resources we enjoyed but that were not available to Albert
Philippot during the previous restoration campaign in 19500–51.17 The many hours
that the conservators (some of whom are also art historians) spent treating the
panels allowed them to observe the work in the closest possible detail, cleared
of substantial overpaints, and to make discoveries – some anticipated in earlier
studies, others hitherto unsuspected. This study has benefi ted from the comments
and observations of all the conservators who took part in the treatment of the
four paintings: Griet Steyaert for the Knights of Christ; Nathalie Laquière, Laure
Mortiaux, Marie Postec, Hélène Dubois and Cécile de Boulard for the Adoration
of the Lamb; Bart Devolder and Kathleen Froyen for the Hermits; and Françoise
Rosier for the Pilgrims. The Just Judges panel, which was stolen in 1934 and never
recovered, obviously did not form part of the study, although we do still have the
original frame.18 Isabel Bedos and Marta Darowska studied the frames and Jean-
Albert Glatigny the supports.
The observations recorded by each conservator are very important, but they
would have been incomplete and insuffi cient without the intensive collaboration of
numerous specialists. Scientifi c imaging is an inexhaustible source of information
that needs to be compared with direct observation of the works. The contribution
of each participant has therefore been crucial: that of the photographers who made
an ongoing visual record of the work as a whole and its details in normal, raking,
ultraviolet and infrared light, as well as the specialists who produced X-radiographs
and infrared refl ectograms before, during and after treatment. Dialogue with the
laboratory may provide support for hypotheses or raise new questions. kik-irpa
scientists paid heed to the conservators’ requests while in turn suggesting analyses
with the potential to deliver fresh information. The same went for their colleagues at
Belgian universities – primarily Antwerp University, but those in Ghent and Louvain-
3
INTRODUCTION
la Neuve, too – and at institutions abroad, such as the National Gallery, London,
the Doerner Institute, Munich, and the Kunsthistorisches Museum, Vienna, all of
which complemented the analyses carried out at kik-irpa. The discoveries were
further facilitated by new and much more effective analytical methods, including
Macro X-ray Fluorescence scans (ma-xrf) and Macro X-ray Powder Diffraction (ma-
xrpd).19 We were likewise able to call on much better microscopes than those used
in the twentieth century, not to mention far more advanced analysis of stratigraphic
sections. The resolution of the infrared refl ectograms is considerably higher now
and the images can be digitally assembled to create ensembles that are easier to
interpret. Macrophotographs are much higher resolution too – in both daylight
and infrared – while X-radiographs are digitally assembled and the interference of
the cradles is reduced nowadays. At the end of the day, everyone involved relied on
the expertise of their peers – conservators-restorers, scientists and art historians – to
place their individual analyses in a broader context. The expert commission, which
has been overseeing this large-scale restoration campaign since 20012, provided
valuable advice, as did our other partners in Belgium and abroad.
Not only have we benefi ted from an abundance of scientifi c imagery, the vast
majority of the documents obtained can be readily consulted and compared
online at the ‘Closer to Van Eyck’ website.20 What is more, thanks to the Van
Eyck Research in OpeN Access (verona) project,21 this documentation can also be
compared on the same site with all the works signed and dated by Jan van Eyck
as well as those generally attributed to him, all documented by the same team of
specialists and according to an identical protocol. Direct contact with the paintings
themselves, however, remains an essential approach in terms of understanding the
creative process of a work made several centuries ago, and even more so for the
Ghent Altarpiece, which had a complex history even before its completion in 1432.22
As was the case with the exterior, treating the paintings from the lower register of
the open altarpiece took a good deal of time: the time spent of examining the panels
prior to each signifi cant decision, the time taken for formulating and answering
questions, and, of course, the time needed for each phase of the restoration process
proper. No matter how long these treatment phases might seem – four years for
the reverse sides of the shutters, three years for the lower register of the altarpiece
interior – they turned out to be far from a luxury. On the contrary, in fact, the
timing was actually quite tight. The time needed to raise and answer questions
is rarely mentioned in the literature on restoration treatments, yet it is extremely
important. The conservators inevitably accumulate a mass of documentation
that then needs to be collated in order to create a coherent account. Each person
records visual data, keeps notes on and photographs the things that they observe,
admire or do not understand right away. Ideally, two or three additional days
ought to be assigned for each day of restoration work, to allow the recorded data
to be classifi ed, analysed and linked to colleagues’ observations and for carefully
reasoned hypotheses to be developed. Regrettably, few projects allow such scope,
so it has been a privilege to have been granted it by the kik-irpa to prepare this
publication. Several members of the kik-irpa team, in collaboration with the
scientists from Antwerp University,23 involved with this second phase of the study
and treatment of the Ghent Altarpiece, took advantage of this opportunity to collate
4
INTRODUCTION
all the observations made by the team, the photographs, the scientifi c imaging and
the laboratory analyses, in order to study the question in more depth and propose
new hypotheses.
The chemical analysis of the materials likewise entailed a degree of interpretation
of the results based on a better understanding of pigment–binder interactions. This,
too, required time and objective consideration. The laboratory results presented in
chapter 3 highlight the degree to which fresh research and new analysis methods
can cast doubt on data acquired in the past. They also show how vital it is in a study
of this kind to contextualize our ever-increasing knowledge of the nature of the
materials used in a pictorial layer. The relatively recent study of the additives used
to modify the properties of the paint layer holds out the prospect of fascinating
revelations concerning the formulation of the materials by these peerless artists.
The many hours needed to treat the paintings gave the conservators the
opportunity to observe every millimetre of the painted surfaces. This work – done
to a large degree via stereomicroscopes that allow the successive layers and their
specifi c material characteristics to be distinguished – offers a unique understanding
of the studied paintings. The present volume proposes a status quaestionis while also
setting out fresh benchmarks, notably a clearer hypothesis as to how far advanced
the paintings were on Hubert’s death and what changes were made by Jan. The
latter were formal in character, but iconographical too, most notably in the addition
of the fountain in the Adoration of the Lamb panel and changes involving the dove of
the Holy Spirit. Art historians will be certain to explore these new data. As such,
the fi nal-phase conservation and restoration treatment of the upper register of the
open altarpiece will provide further, unprecedented information that will complete
our knowledge of this prestigious ensemble and most likely answer some of the
questions raised here.
Notes
1 Coremans 1953. The project very much 4 Dubois 20018a; Dubois 200200a; Sanyova
planted the seed for the characteristic et al. 200200.
interdisciplinary work of kik-irpa. 5 Van Grevenstein et al. 20011. Regarding
Claes et al. 20019. the progress of the project up to 20016,
2 Coremans 1953; Van Asperen de see Martens et al. 200200a.
Boer 1979. Without attempting to
be exhaustive, we will refer to other 6 We are grateful to the staff of the msk
publications on the pictorial technique and its succeeding directors, for their
of the Ghent Altarpiece in the notes to the constant support.
chapters that follow. 7 Currie et al. 20017.
3 The term surpeint (overpaint) is 8 Fraiture 20017.
nevertheless used widely in L’Agneau 9 Panofsky 1953, pp. 2007–22. Panofsky
mystique au laboratoire, without always
refers to earlier authors in note 2008/3.
being clearly defi ned. Coremans 1953,
see e.g. p. 19. 10 Fransen, Stroo 200200.
5
INTRODUCTION
6
1
Introduction
This chapter describes the methodology and results of the conservation and
restoration of the paint surface of the Knights of Christ, the Adoration of the Lamb,
the Hermits and the Pilgrims (fi gs. 1.1 and 1.2).1
The experience acquired during the restoration of the Ghent Altarpiece’s exterior
led us to expect several layers of degraded varnish and restorations in the phase
discussed here. These layers obscured the colour values, the subtle play of light
and textures, and the spatial illusion of the original, Eyckian paint surface. Also,
substantial lifting of small paint fl akes could not be treated without removing
these non original layers.
The restorers were naturally aware of the treatment by Albert Philippot in
1 950
0–51, and of other restorations carried out in the nineteenth century, but their
extent and location were not mapped precisely. Nor were they surprised to discover
old overpaintings similar to those encountered on the closed polyptych, such as
mid-sixteenth-century overpainting and, possibly, even older interventions that
were completely hidden.2 As described (albeit not very clearly) in Paul Coremans’s
L’Agneau mystique au laboratoire, Philippot removed some of the nineteenth-century
restorations as well as some of those carried out in the sixteenth century (fi gs. 1.3
and 1.4).
Coremans and his team concluded that zones of the Adoration of the Lamb had
been overpainted in the sixteenth century.3 Their deduction was based on the
pigment analysis of paint samples, on the examination of the paintings during
their restoration, and on the fact that the overpainted areas were closely followed
(facing page)
Adoration of the
by Michiel Coxcie (1499–1592) in his copy of the Ghent Altarpiece of 1557–58.4
Lamb: detail of the Although a re-examination of the cross-sections taken during the 19500–51 campaign
drapery of one of seemed to refute Coremans’s assumptions,5 his initial interpretation was eventually
the virgins during confi rmed during the present restoration project.6 Coremans also suggested that
overpaint removal Jan van Scorel of Utrecht (1495–1562) and Lancelot Blondeel of Bruges (1498–
(see also fi g. 1.16). 1 561) were the authors of these old overpaint.7
9
I. THE CONSERVATION-RESTORATION
1 .1
1.2
10
I. THE CONSERVATION-RESTORATION
1 .3
11
I. THE CONSERVATION-RESTORATION
1 .5a
1 .5b
1.5c
12
I. THE CONSERVATION-RESTORATION
Treatment procedures
Varnishes composed of synthetic resins applied since the 19500s were fi rst removed.
These were followed by uneven layers and residues of older, darkened natural resin
and oil varnishes (fi gs. 1.7a and 1.9a–b), as well as local, thick and discoloured
retouchings associated with these varnishing campaigns. This approach, already
adopted during the fi rst phase of the restoration, allowed a progressive insight into
past interventions. There were only a few nineteenth-century retouchings on the
shutters, but on the Adoration of the Lamb panel many restored areas thickly painted
with pigments that did not come into use until the nineteenth century, such as
cobalt blue, or chrome oxide green, spilled onto the original paint. Particularly
crude and darkened were retouchings in the foreground fi gures (fi g. 1.10). Their
13
I. THE CONSERVATION-RESTORATION
1 .7a
14
I. THE CONSERVATION-RESTORATION
1 .7b
15
I. THE CONSERVATION-RESTORATION
1.9a 1.9b
16
I. THE CONSERVATION-RESTORATION
1. 10
1. 11 17
I. THE CONSERVATION-RESTORATION
Fig. 1.12a–d.
Diagram of the
extent of sixteenth-
century overpainting
(in red).
1 .12d
18
I. THE CONSERVATION-RESTORATION
Fig. 1.13a–d.
Diagram of the
actual paint losses
that were hidden by
sixteenth-century
overpaints (in red).
1.1 3d
19
I. THE CONSERVATION-RESTORATION
Fig. 1.14a–b.
Adoration of the
Lamb, the tower of
Utrecht cathedral,
before (a) and
after (b) overpaint
removal, during
restoration.
1 .1 4a 1.1 4b
1. 15
20
I. THE CONSERVATION-RESTORATION
1. 16
1. 17
21
I. THE CONSERVATION-RESTORATION
22
I. THE CONSERVATION-RESTORATION
1. 19
Fig. 1.200a–b.
Adoration of the
Lamb, detail.
Face of one of the
virgins before
(a) and after (b)
the removal of
overpaint.
1.20
0a 1 .20
0b
23
I. THE CONSERVATION-RESTORATION
1.21 b
1 .21c
24
I. THE CONSERVATION-RESTORATION
1.22a 1 .22b
25
I. THE CONSERVATION-RESTORATION
1 .23a
26
I. THE CONSERVATION-RESTORATION
1.24
27
I. THE CONSERVATION-RESTORATION
surface. The overpaint, varying in texture, hardness and thickness, was removed
where it could be confi dently identifi ed and only where this could be done safely.
The fi nalization of the cleaning process included the removal of remnants of
putty spread over the original surface, as well as thick, browned and uneven old
varnish layers that had accumulated in the darker areas, particularly on the wings.
This last step has further revealed the luminosity of the painted surface, the depth
of the landscapes and the almost tactile appearance of painted metals, stones, plants
and fabrics (fi g. 1.24).26
The restoration treatment resulted in a major retrieval of areas of original
paint, since the amount of actual paint loss preceding the mid-sixteenth-century
restoration (about 5% in the Adoration of the Lamb, and even less on the wings)
was only a fraction of the overpainted surface. Paint losses were caused by the
cumulative effect of the natural degradation of materials, human intervention (such
as opening and closing of the altarpiece and manipulations relating to the liturgy),
accidental impacts and inept cleaning and restorations. Paint layer analysis suggests
that, just as with the paintings of the exterior, the mid-sixteenth-century restorers
started with a superfi cial cleaning. They did not remove all older varnishes or
earlier restorations before they set about overpainting.
During the overpaint removal we have become aware of even older, carefully
painted additions applied directly to the Van Eycks’ paint surface. These additions
were not removed because, at the time, they could not be confi dently identifi ed.
Further research was needed and the fi rst results thereof are described in this
volume. It is not known whether the author(s) of these additions also worked on the
exterior of the shutters: diverse types of restoration were found there under the mid-
sixteenth-century overpainting and were removed when this could be done safely.27
28
I. THE CONSERVATION-RESTORATION
Fig. 1.26.
Attributed to Jan
van Eyck (and
workshop assistant),
The Crucifixion
(detail), New York,
The Metropolitan
Museum of Art,
acc. 33.92ab.
1. 26
29
I. THE CONSERVATION-RESTORATION
surface after retouching. This coating, giving a soft, even sheen to all the paintings
without masking their delicate textures, concluded the treatment.
As with the fi rst phase of the restoration, the removal of the sixteenth-century
overpaint from the lower register of the interior allowed the astounding recovery
of the Van Eyck brothers’ original paint surface and contributed to securing its
conservation for future generations. The treatment of the upper interior panels –
the third phase of the restoration – will no doubt lead to further revelations on this
unique masterpiece.
Notes
1 Pending future research on the supports, 7 See note 3, above. Dubois (20018a,
treated by Jean-Albert Glatigny, and pp. 756–600, and 200200a, pp. 23–29)
the frames, treated by Marta Darowska argued that the campaign may have
and Isabel Bedos, see ‘Closer to Van been motivated by the context of early
Eyck’ (http://closertovaneyck.kikirpa. Counter-Reformation initiatives in the
be); Froyen, Dubois 200200, pp. 28–37 orbit of the Habsburg court.
and 216–39. Materials and methods used 8 Dubois 20018b.
for this second phase of the restoration
are identical to those of the fi rst phase: 9 De Schryver, Marijnissen 1953, n. 54,
Augustyniak et al. 20017 and Depuydt- pp. 500–52 and n. 65, pp. 55–56.
Elbaum et al. 200200. The twentieth- 10 Stehr, Dubois 20014.
century copy of the stolen Just Judges 11 This approach was supported by the
was not restored. It was conserved in International Commission advising on
20010 (Van Grevenstein et al. 20011, the treatment at the time.
pp. 189–99), and its frame, which was
not stolen, was restored during the 12 These rays were removed during the
present campaign. present restoration.
2 On the discovery of old overpaint on the 13 See also fi g. 1.3.
exterior, see Postec et al. 20016; Depuydt- 14 Coremans (1953, pp. 114–15) thought
Elbaum et al. 200200, pp. 124–35; Martens that the lamb was overpainted after being
20015; Van Grevenstein 20015. damaged in the fi re that occurred in
3 Coremans, Loose, Thissen 1953, 1 822.
pp. 98–99 and 101–17. Old overpaints 15 Panofsky 1953, pp. 218–19; Coremans
were also reported on the Deisis and on 1 953, p. 108; Van Asperen de Boer 1979,
the Angels in the upper register. pp. 195–96 and 200 00–2001.
4 On Coxcie’s copying approach, 16 Coremans 1953, pp. 109 and 112.
see Dubois 20017. Van Scorel was canon of St Mary’s in
5 Brinkman et al. 1988/89, pp. 35–37; Utrecht: Faries 1997, p. 107.
Van Asperen de Boer (1979, pp. 155–63 17 Philippot, Sneyers 1953, p. 83;
and 172–78) was not convinced that the Coremans, Loose, Thissen 1953, pp. 107,
evidence presented by Coremans proved 11 5 and pls. xxx, xxxi and xl.
that large sections of the interior upper
register were overpainted. 18 Cross-section showing the overpaint on
6 Depuydt-Elbaum et al. 200200, top of thick varnish; see fi g. 3.15b, cross-
pp. 147–151. On the analytical detection section 11/12 and fi g. 2.11d, cross-section
1 0/73.
of overpaint on the exterior, see Van der
Snickt et al. 20017; Sanyova et al. 200200. 19 In spite of a large loss in the tower, its
Coremans and his team had not noticed colour and proportions and the shape of
that the exterior of the shutters was also its buttresses provided clues towards its
overpainted. identifi cation.
30
I. THE CONSERVATION-RESTORATION
20 UGent professors Marc Boone, Anne- 26 These results are best observed on
Laure van Bruaene, Jan Dumolyn the website ‘Closer to Van Eyck’:
and Maximiliaan Martens, urban http://closertovaneyck.kikirpa.be.
archaeologist Marie-Christine Laleman 27 Depuydt-Elbaum et al. 200200,
and architectural historian Luc pp. 138–39. Older overpaint and zones
Devliegher advised on the identifi cation of questionable authorship on the exterior
of the buildings. are mapped in Augustyniak et al. 20017,
21 Philippot may have removed old pp. 225–26.
overpaint from the dove when he thinned 28 All materials used for fi lling and
down nineteenth-century restorations retouching are stable and reversible
(Coremans 1953, p. 108). without causing any damage to the
22 Van der Snickt et al. 200200. original; Depuydt-Elbaum et al. 200200,
23 Danny Praet (UGent) and Barbara Baert pp. 152–56.
(KULeuven). 29 See note 26.
24 Froyen, Dubois 200200, pp. 74–77. 30 See note 200.
25 Van der Snickt et al. 200200. See also
several examples of this documentation
in this volume.
31
2
The Van Eycks’ Creative Process and the Different
Stages in the Execution of the Interior Lower
Register of the Ghent Altarpiece
Introduction
33
2. THE VAN EYCKS’ CREATIVE PROCESS
for our study to follow the approach suggested by these authors, with analysis of
the paint layers by zones or by areas of representation (e.g. vegetation, architecture,
fi gures). To avoid swamping the reader with details, we will begin by outlining
the new hypotheses that are presented here. It should be noted that these are based
not only on the technical and scientifi c studies and laboratory analyses, but also
on the stylistic study discussed in chapter 4. The hypotheses we present may be
introduced with reference to the central panel.
34
2. THE VAN EYCKS’ CREATIVE PROCESS
developed the fl ora by painting a wide variety of the shutters were painted with a mixture of lead
plants and applied at least one new layer of azurite white and azurite, the top layer in the central
to the sky. Jan reworked the urban landscapes on panel comprises lead white and ultramarine. This
the horizon, adding a few buildings, including ultramarine-based layer and most of the blue hills
the tower of Utrecht cathedral, the former currently visible in the distance in the central
Abbey of St Bavo’s and a tower reminiscent of panel were created during the same intervention,
Our Lady’s Church in Bruges. To achieve this, and clearly after completion of the second-stage
he concealed the initial composition, while also buildings, which the ultramarine-based layer
retaining certain fi rst-stage elements, which were overlaps slightly. The presence of this latter
incorporated into the fi nal version. This makes it stratum in the central panel modifi es the overall
more complicated to distinguish clearly between tonality of the sky relative to that of the shutters.
the elements belonging to each of the fi rst two All this inclines us to question the originality
stages of development. It might also have been of the ultramarine sky, although it undoubtedly
Jan, lastly, who painted a fi rst dove surrounded belongs to the zone with the halo that is now
by a coloured halo, thus abandoning his brother’s visible around the dove. The halo was painted in
project with the rays. two layers, however, so it is not impossible that
Jan painted a fi rst halo, with or without the dove.
The dove we see today can hardly be attributed to
Third stage him, but it, too, was painted in at least two layers.
While we cannot exclude the possibility that
It appears that Jan was not the only person – or this was the work of one or more assistants, the
the last – to have reworked the lower register interventions in question seem to post-date the
of the altarpiece interior. Numerous retouches, Van Eycks. Even though we think that these
invariably on the surface and scattered more interventions did not form part of the altarpiece’s
or less across the paintings studied here, were creative process, they are bound to come up
applied to zones we believe to have been painted frequently. Considering – and also localizing
and completed by Jan, but instead of enhancing – them is important if we are to appreciate the
them, they made them heavier. Van Eycks’ technique and attempt to distinguish
The fi nal paint layer in the sky of the central between the contributions of the two brothers.
panel also raises questions: whereas the skies in
35
2. THE VAN EYCKS’ CREATIVE PROCESS
The panels in the lower register of the altarpiece interior are made of Baltic oak – four
planks laid horizontally for the central panel and three vertical planks for each wing.11
Preparatory layers
Ground
Both the composition and application of the ground do not seem to differ from
those found on the altarpiece exterior. The ground layer – a mixture of chalk and
animal glue – was applied in several layers12 to the already framed panels and hence
also to the mouldings of the frames. The presence of unpainted borders and barbs
on each panel attests to this.
As in the case of the paintings of the altarpiece exterior, brushes and a spatula
were used to apply the ground.13 Few bristles from the brushes were found in
this instance, however (fi g. 2.1),14 unlike the paintings of the closed polyptych, in
which they were particularly numerous and mostly trapped in the ground layer.15
Such bristles as were found were essentially located in the pictorial layer. Bristles
trapped in the ground layer might be less obvious in the paintings studied here,
36
2. THE VAN EYCKS’ CREATIVE PROCESS
2.2a 2.2b
Fig. 2.2a–b. Hermits,
detail (a); striations
indicating the use of
a spatula in applying
given the larger number of paint layers.16 A few striations demonstrating the use of
the ground, irr a spatula have been identifi ed, in the Hermits, for instance, along the right edge in
detail (b). the drapery of the tall, bald fi gure (fi g. 2.2).
A strip of canvas was glued to a section of a joint in the central panel.17 It was
noted while restoring a loss (fi g. 2.3) and can be seen in the infrared refl ectogram
and X-radiograph.18 The strip, measuring approximately 28 by 4 cm, was glued
in place before the ground was applied. Rather than being centred over the
joint, it runs roughly 1 cm above it but approximately 3 cm below it. Although
reinforcement of this kind has been found in a few pre-Eyckian works,19 it becomes
rarer in the course of the fi fteenth century. In the Virgin of Canon Van der Paele,
however, plant fi bres were glued to the obverse side over the joints of both the
panel and the frame.20 The question in the Adoration is whether the canvas strip
was intended to reinforce a weak point in the panel or to conceal some alteration in
the wood.21 Or might it have been a means of evening out a local difference in level
between two planks?22
Underdrawing
The infrared photograph and refl ectogram allow us to make out lines indicating
the contours of the forms and the hatching of the modelling belonging to the
underdrawing, but as we will see, in many places it is not the drawing that is legible
in these documents but the fi rst, already painted stage, which is now concealed.
37
2. THE VAN EYCKS’ CREATIVE PROCESS
+/- 28 cm
2.3a
Fig. 2.3a–b. Adoration of the Lamb, detail, diagram: strip of
canvas (indicated by yellow frame) glued to a section of a
joint (dotted line), under the ground layer (a) (see also same
detail in irr and xr on ‘Closer to Van Eyck’ website); canvas
visible during restoration in a lacuna (b).
38
2. THE VAN EYCKS’ CREATIVE PROCESS
The study of the paintings on the exterior of the altarpiece identifi ed different
stages in the development of the drawing as well as different media – in this case
a dry medium for the fi rst laying out of the main elements of the composition,
followed by a liquid medium for the more elaborate modelling of the forms, using
lines and hatching of varying thickness, but also washes.
We also think that a fi rst drawing stage using a dry medium can be seen in the
lower register of the altarpiece interior which, as in the panels of the exterior, is
rarely discernible and then only in locations where it was not followed by the liquid
drawing. In the central panel, this is the case with a few of the foreground fi gures.
The folds in the drapery of the prophet in the blue cloak differ from those actually
painted and, above all, a rectangular shape – possibly a book that was not painted
in the fi nal composition – traverses the fi gure (fi g. 2.5). A fairly similar form is
Fig. 2.4a–b. found in one of the prophets kneeling in the foreground and seen from behind
Adoration of the (cve–5). These few lines visible in the infrared refl ectogram seemingly display the
Lamb, detail (a); characteristics of a dry medium, namely a discontinuous or even grainy line. They
underdrawing with defi ne summary shapes, without hatching for volume. Lines like this made using
nervous strokes, short a dry medium do not show in the infrared refl ectograms of the wing panels of the
lines drawn with a lower register of the exterior, although this does not mean they do not exist.26
certain verve, often As in the panels of the altarpiece exterior,27 most of the underdrawing visible
curved, crossing here in the infrared refl ectogram seems to have been done in a brush loaded with an
the long lines
ink or a fl uid paint. A few droplets at the end of lines in the underdrawing (fi g. 2.6)
demarcating the folds
(irr detail) (b).
2.4a 2.4b
39
2. THE VAN EYCKS’ CREATIVE PROCESS
2.5a 2.5b
2.6
40
2. THE VAN EYCKS’ CREATIVE PROCESS
41
2. THE VAN EYCKS’ CREATIVE PROCESS
material (containing lead white) fi lled the hollows left by these incisions. There
are approximately 28 of these rays,29 which do not correspond with the 129 gilded
rays we see now. The latter also converge towards a point located near the barb of
the paint layer, approximately 1 cm below the convergence point of the drawn and
scored rays, which is located outside the limits of the panel. A small nail hole was
located near the barb (fi g. 2.10), as if a cord had been attached there to serve as a
guide for drawing the future gilded rays. There is a certain concordance, therefore,
between the drawn and the scored rays, but a substantial difference compared with
the gilded rays. The latter point will be discussed further below.
The lines were probably scored using a stylus and a ruler to ensure they would
be straight and remain visible, given that they would not be executed until after
completion of all the pictorial layers, which would have obscured the black drawing.30
It should be noted that neither the scored nor the drawn rays were discontinued
to leave space for the dove, for which no preparatory drawing is visible in the
infrared refl ectogram; this has prompted some authors to conclude that it was not
initially planned.31
Other scored lines can be detected in large numbers at the fountain, but they
were made at a later stage: they were not incised in the ground32 but in the green Fig. 2.9a–b.
pictorial layer on which the fountain is painted (fi gs. 2.27 and 2.28). We will Adoration of the
return to this later. Van Asperen de Boer already noted in 1979 that no preparatory Lamb, detail (a);
drawing could be detected below the fountain,33 and this is confi rmed by the new washes applied in
infrared refl ectogram. broad brushstrokes
(red arrow, irr
detail) (b).
2.9a 2.9b
42
2. THE VAN EYCKS’ CREATIVE PROCESS
43
2. THE VAN EYCKS’ CREATIVE PROCESS
0
10 9 10
0
9
8 8
7 7
6
6
5
5
4
4 3 2
3 2
1
1
50
0 μm 50 μm
2.11d 2.11e
44
2. THE VAN EYCKS’ CREATIVE PROCESS
7 7
6 6
5 5
4 4
2 3 2 3
1
1
50
0 μm 50
0 μm
2.12d 2.12e
45
2. THE VAN EYCKS’ CREATIVE PROCESS
2.13a
20 μm
2.13b 2.13c
Fig. 2.13a–c.
Adoration of the
having been broken, bent to follow the indentation of the incision. The second Lamb: location of
layer (layer 4 in fi gs. 2.11 and 2.12), itself translucent but more pigmented, is sampling (a–b);
essentially composed of lead white, calcium carbonate, carbon black and a few cross-section 10/69
red or ochre particles, giving it a warm tonality. This is most likely the priming, vis, incision made
also called imprimatura, previously detected below the paintings of the altarpiece in the ground
exterior and which corresponds with the pigmented intermediate layer that is very layer to mark the
often present in medieval pictures painted in oil.41 position of the
rays surrounding
the dove.
Paint layers The stylus used to
score the drawing
also cut across
Meadow and fountain the translucent
isolation layer
As Van Asperen de Boer already noted, not only has no preparatory drawing been (white arrow) (c).
detected beneath the fountain in the infrared refl ectogram, the fountain was also
painted over a layer of green paint42 – basin and shaft alike. The recent research and
analysis show that there was an initial project involving a different meadow with
what we believe to be a spring at the centre. The stone fountain was added later and
also appears to have been designed in two stages.
First stage ma-xrf analysis (fi g. 2.14a-b) detected copper beneath the current
fountain, suggesting a painted meadow.43 This could also be made out during the
restoration process in losses or abrasions in the stone basin (fi g. 2.17). Microscopic
observation and cross-section analysis identifi ed a stratigraphy comprising several
green layers including glazes.44 Layering of this kind is traditional in Flemish
painting,45 with relatively light and opaque base layers consisting of verdigris
and more opaque pigments such as lead white and/or lead-tin yellow, and a glaze
46
2. THE VAN EYCKS’ CREATIVE PROCESS
consisting essentially of verdigris. The ma-xrf copper map also reveals a central
area with almost no copper (black in the ma-xrf copper map) in the shape of
an oval, the lower part of which shows a bulge towards the bottom of the panel
(fi g. 2.14a). This is a reserve, an area that was left blank in the meadow46 (fi g. 2.14b)
and then painted black. The black, ovaloid form can be made out in the infrared
refl ectogram (fi g. 2.14c–d). The black layer is also visible through abrasions in the
pictorial layer of the stone basin, which was painted over it (fi g. 2.18). The edges of
the black oval shape overlap the edges of the green layer (visible in raking light) of
the surrounding meadow (fi gs. 2.18 and 2.14e–f).
So what might this shape with its bulge at the bottom have represented? A
manuscript illumination with God Placing Adam in the Garden of Eden by the Master
of Berry’s Clères Femmes, dating from around 14005 (fi g. 2.15), appears to provide
an answer.47 It features two dark springs, square in shape here, with a similar bulge
along their lower edges, from which the stream of water fl ows. These dark shapes
were clearly painted over the surrounding grass. The dark tone evokes the depth of
the springs, while the superimposition of the dark tone on the light green of the
meadow – at the far end of the springs – suggests their edges and the transparency
of the water. Comparison with this miniature suggests that the black oval in the
Adoration served as the ground tone for the rendering of the water of the spring that
was planned at that point. Although black, the paint – applied in large brushstrokes
(visible in the irr) – probably displayed a variety of shades according to the action
of the brush that moved around more or less paint and so created variations in
translucency. The same technique may be found in the Meeting of the Magi in the
Très Riches Heures du duc de Berry (fi g. 2.16),48 where, once again, a dark layer overlaps
onto the light tone of the surrounding rocky plateau to suggest the borders of the
pools and water overfl owing.
A sample taken from the refl ection of the shaft of the fountain in the Adoration
(fi g. 2.19) allows this black layer – itself covered with light-green paint49 before the
brown paint of the shaft was added – to be observed. The light-green layer most likely
represents water or some ripples. The spring thus appears to have been painted to
an advanced degree. The surrounding meadow, consisting of several layers including
a glaze, similarly suggests that the composition at the centre of the Adoration panel
was well advanced. The documents at our disposal50 do not allow us to verify whether
a fl ow of water, originating at the spring, was painted. However, the initially painted
shape with the bulge at the bottom and its comparison with the springs in the Bible
historiale suggest that a ribbon of water of this kind was intended.
Within the overall composition of the polyptych, this presumed spring is located
at the centre of the meadow zone, surrounded by all the groups of fi gures and
beneath the altar. It is integrated in a perspective that gives full meaning to the
initially planned shape with the bulge.51
Second stage It was decided at some point to turn this natural spring into an
octagonal stone basin.52 As noted, the fountain was painted over a meadow. In
reality, the entire fi eld in the central panel was repainted at this stage in two layers
– fi rst with a mixture giving a less translucent appearance than the fi nal glaze of
the fi rst meadow, namely verdigris, lead white and lead-tin yellow, then with a
47
2. THE VAN EYCKS’ CREATIVE PROCESS
2.14a 2.14b
2.14c 2.14d
2.14e 2.14f
Fig. 2.14a–f. Adoration of the Lamb, the fountain with fi lled losses during
restoration. Different steps in the realization of the hypothetical spring of
the fi rst stage. ma-xrf Cu-K map (a); diagram showing green layer of the
fi rst painted meadow, with oval shape left in reserve, visible in ma-xrf
Cu-K map (b); irr (c); diagram of the black, ovaloid form that can be
made out in the irr (d); diagram vis with the overlapping outlines of the
reserve (green) and the black, ovaloid form (black) (e); diagram layering
the black, ovaloid form on top of the green and the reserve (f).
48
2. THE VAN EYCKS’ CREATIVE PROCESS
2.15a 2.15b
Fig. 2.15a–b. Three springs in God Placing
Adam in the Garden of Eden, miniature in Guyart
des Moulins, Bible historiale, c. 14005, Paris,
Bibliothèque de l’Arsenal MS 50057, fol. 8r.
2.16a 2.16b
Fig. 2.16 a–b. Two pools with reed tufts in the
Meeting of the Magi, miniature in the Limbourg
Brothers, Les Très Riches Heures du duc de Berry,
Chantilly, Musée Condé, MS 65, fol. 51.
49
2. THE VAN EYCKS’ CREATIVE PROCESS
2.17a 2.17b
2.17c 2.17d
50
2. THE VAN EYCKS’ CREATIVE PROCESS
2.18a 2.18b
2.18c
51
2. THE VAN EYCKS’ CREATIVE PROCESS
translucent layer based on verdigris. This stratigraphy can clearly be seen in the
paint samples taken from these locations. During the restoration treatment, the
glaze of the fi rst meadow was also detected in several losses (fi g. 2.200).
The painter of this second stage did, however, use the glaze of the fi rst meadow
here and there to enliven the tufts of grass and the fl owers he painted, although
this did not prevent him from adding localized glazes himself, to indicate a small
leaf here, a stem there (fi g. 2.21). For their part, the fl owers are painted by means
of swiftly applied touches which, when viewed under the microscope, might give
the impression that they are merely coloured patches resulting from the painter’s
playful handling of paint, but which actually represent perfectly identifi able
varieties.53 This is the case, for instance, for the rose campion (fi g. 2.22) at the foot
of the pomegranate tree to the left of the confessors. This was done by applying a
preliminary red layer to which, once dry, the shades of the fl ower were added with
touches of pure white and with red glazes, which were applied wet-in-wet, causing
them to blend. Working wet-in-wet – blending the colours directly on the panel
rather than the palette – allows layers that are not dry yet to be combined without
totally mixing them. The touch has to be swift and precise, but it arises from a
clear idea of the desired result and a perfect mastery of the gesture, because no Fig. 2.19a–c.
reworking is possible without risking the loss of the intended effect, namely the Adoration of the
creation of visual impressions rather than accurate lines and shapes. The painter’s Lamb: location of
virtuosity is apparent in the splitting of blades of grass using the back of the brush sampling (a–b);
in the fresh paint (cve-3) or dabbing the glaze with the fi ngertips, leaving visible cross-section 10/83
(c): preparatory
prints (fi g. 2.23).
layers (1, 2, 3); the
This second meadow skirts around several groups of fi gures, namely those black layer (4); a
with the large ones in the foreground, but also the more distant groups with the green layer (5);
confessors and virgin martyrs, as well as the altar, the angels and the lamb, which two brown layers
were probably already fi nished, or at least well advanced. We will return to the corresponding to
latter in due course. This skirting of the forms is clearly noticeable in places by a the refl ection of
ridge in the paint surface, marking the limits of the second meadow. This is the case the shaft in the
around the fi gures of the prophets and the apostles on either side of the fountain water of the basin
(fi g. 2.24). In some places, however – below the virgin martyrs, for instance – the (6 and 7); non-
transition between the two stages appears gentler. The fi rst meadow is still partly original varnishes
and retouches now
removed (8, 9, 10).
10
0
7 8-9
6
4
3
2
50
0 μm
52
2. THE VAN EYCKS’ CREATIVE PROCESS
Fig. 2.200a–b.
Adoration of the
Lamb, detail: green
glaze of the fi rst
meadow visible in
the losses of the
second stage.
2.200a 2.200b
visible at the interface between fi gure and meadow (fi g. 2.25), suggesting that the
second-stage meadow stops just below the virgin martyrs group which, if not fully
painted, will have been well advanced at least.
It is in the green layers of the second, freshly painted meadow that the contours
of the octagonal fountain were traced by scoring (fi gs. 2.26 and 2.27).54 The
fountain was not drawn on the ground layer, as this had already been covered by
the pictorial layers of the fi rst stage. The scoring in question is multiple, indicating
a search for the form of the basin, in the same way that a drawing would be subject
to adjustments. It even appears that guide marks were drawn, possibly using a
compass (fi g. 2.28).55
The stone basin of the fountain was painted in two principal layers: a preliminary
grey layer (apparently fl at and uniform), followed by a second, more modulated
grey-ochre layer along the more or less illuminated faces of the basin (fi g. 2.29).56
Veins or irregularities in the stone were suggested by translucent brown strokes.
The metal column was lower at this stage than in the fi nal version. The fi nal
design of the fountain consists of a central metal column decorated with protruding
rings spaced along the height of the shaft, then two series of dragon-like gargoyles
spitting water. The fi nal series of gargoyles ranged around a metal sphere is
surmounted by the fi gure of an angel. It appears that the column was not initially
intended to rise beyond the second ring, since the section below it was painted
53
2. THE VAN EYCKS’ CREATIVE PROCESS
Fig. 2.21a–c.
onto the new meadow without grass,57 while the section above it was executed over Adoration of the
already painted grass, as shown in the infrared photograph (fi g. 2.300). The column Lamb, detail (a);
was thus apparently not going to be so high and the shaft would have been simpler green glazes from
and would only have measured half its current height. Even so, the idea of a taller the fi rst stage left
column probably emerged quickly, since the stylistic and technical characteristics uncovered, visible
appear to be the same for its full height. Lines were also scored to mark the limits of on the surface (white
the enlargement of the fountain’s section. The water with the gemstones at the foot arrow), and localized
of the basin may likewise have been enlarged at this point, as it is partially painted green glaze added
over the grass in the same way as the upper section of the column. in the second stage
The column was painted over a brown base layer without as yet paying attention (red arrow); vis (b);
raking light (c).
to the rings, even where the shaft has been extended over the meadow for which
grass had already been painted. Dark brown, light yellow and ochre touches were
then applied to shape the rings and simulate light refl ecting on the metal.58
First stage The shutters also feature areas with clearly visible grass on the ground
or on top of the crags and in the uniform green expanses in the distance. For the
most part, however, the soil in the shutters consists mainly of rocky or sandy paths,
although it appears that for certain of these paintings a fi rst stage also existed where
it had vegetation. There are several indications, for example, that the soil beneath
the horses in the Knights of Christ was originally conceived as green. Small paint
losses in the upper layers show an underlying green layer of this kind,59 which has
no reason to be present beneath the brown tonality of the current soil (fi g. 2.31). The
uniform presence of this green layer – most likely based on verdigris – is confi rmed
by the ma-xrf copper map.
54
2. THE VAN EYCKS’ CREATIVE PROCESS
2.22a 2.22b
Fig. 2.22a–b. Adoration of the Lamb, detail:
working wet-in-wet to depict the rose campion.
2.23a 2.23b
Fig. 2.23a–b. Knights of
Christ, detail: fi ngerprints.
55
2. THE VAN EYCKS’ CREATIVE PROCESS
2.24a
2.24c 2.24b
Fig. 2.24. Adoration
of the Lamb, detail (a);
the ridge marking
the limits of the
second meadow skirts
the outlines of the
angels, prophets and
apostles – photograph
in refl ected light (b);
microphotograph (c).
56
2. THE VAN EYCKS’ CREATIVE PROCESS
2.25a 2.25b
Fig. 2.25. Adoration of the Lamb,
detail (a); transition zone between
fi rst-stage (yellow arrow) and
second-stage meadow (white arrow).
2.26a 2.26b
Fig. 2.26a–b. Adoration of the Lamb,
the fountain, diagram of scored lines (a);
scored lines (white arrows) and scored
guide marks, possibly using a compass
(yellow arrow) – photograph taken in
raking light (b).
57
2. THE VAN EYCKS’ CREATIVE PROCESS
2.27a
Fig. 2.27a–b.
Adoration of the Lamb,
detail (a); line incised
in the green paint
layer marking the
position of the basin –
microphotograph (b).
2.27b
Second stage Adjustments to the horses’ legs in the Knights – also reported in previous
studies60 – are contemporary with these changes to the soil. The positioning of the
legs, as we see them now, occurred during the second stage, when the green soil
was changed to brown. A few corresponding indications have also been detected in
the soil of the Pilgrims.61 The possible adjustment of the nature of the soil in this
painting was accompanied by changes in the position of the feet of fi gures behind
the giant St Christopher. Present in a previous stage, they were covered by the tufts
of grass we see now.62 It therefore appears that modifi cations to the soil were also
made in the wings after initial painting.
First stage The vertical crags in the shutters were executed over a brown or
grey base tone. The division into parallel strata was done by juxtaposing lighter
expanses of varying size with brown lines that suggest deep fi ssures in the rock or
slight differences in level. The red surface tones, lastly, were evoked by a few swiftly
applied touches (cve-21).
58
2. THE VAN EYCKS’ CREATIVE PROCESS
2.28a
2.28b
2.29a
59
2. THE VAN EYCKS’ CREATIVE PROCESS
2.300a
2.300b
Where the soils in the wings were originally grassier, the peaks of the crags were
rockier than they are today. The grassy areas at the top of the crags were painted
during the second stage: the rocky passages planned in the fi rst stage were more
arid and lacking in vegetation. The infrared refl ectograms of the Knights of Christ
and the Hermits suggest rocky plateaus below the grassy zones at the top, traversed
in the middle by a change in level in the crag (fi g. 2.32a, cve-16 and cve-22).
In the Knights, a sample taken from the summit of the crag, as well as careful
observation of the left edge (fi g. 2.32c–d), does indeed show a fi rst, greyish-ochre
paint layer, which must have corresponded with the painting of the rocks before
the grass was laid over them.
60
2. THE VAN EYCKS’ CREATIVE PROCESS
The skies in the central panel and the wings were built up through a signifi cant
number of layers: a base of indigo and lead white followed by three layers of azurite
mixed with lead white65 and a fi fth and fi nal layer of ultramarine and lead white
(fi g. 2.45). We will return in a moment to the ultramarine and lead-white layer,
which is only found in the central panel and mostly runs around the edges of the
currently visible architecture.
The sky is painted with a gradation from white horizon to blue upper sky. Existing
samples taken from the sky in each of the four panels show that these shades had
already been introduced as of the fi rst pictorial layer, since the lower part of the sky
was begun solely in lead white and the upper part with a mixture of lead white and
indigo. This progressive variation in colour is confi rmed in the subsequent layers.
We can see this in the ma-xrf copper map, which detects a greater presence of this
element – and hence of azurite – towards the top of the sky in each of the four panels.
Was it necessary to apply four pictorial layers, each repeating the same shades in the
gradation of the sky, to achieve the desired result? The economy of means we fi nd
61
2. THE VAN EYCKS’ CREATIVE PROCESS
in Jan van Eyck’s art66 argues against such complexity, even if no samples have been
taken from the skies in other paintings by him.67 This stratigraphy is more likely
explained by the superimposition of several interventions.
An area of the sky in the Knights shows damage to the paint layer that is not
visible on the surface but beneath it, in the X-radiograph and infrared refl ectogram
(cve-15).68 A joint appears to have opened up, requiring local gluing. The sky in
the zone immediately surrounding this damaged area consists of a fi rst indigo-
based layer followed by three layers of azurite. Investigations carried out during
the present restoration treatment allowed it to be further determined that there is a
lacuna deep in this stratigraphy in the fi rst layer of azurite, which is missing at that
point.69 This degraded zone only includes the fi nal two azurite-based layers. In the
corresponding location on the reverse of the painting, namely in the panel with Joos
Vijd, a network of wider cracks on either side of the joint forms what is probably a
trace of an intervention in the preparatory layers,70 beneath the pictorial layer which
appears to us to be original. In other words, it is a very early problem, which required
the reworking of the sky at a date prior to or contemporary with Jan’s intervention.
This possibly indicates that the fi rst azurite layer belongs to the fi rst stage.71
Study of the central panel helps us to locate the two upper azurite layers. In
certain places, we observe in raking light and with the naked eye a ridge in the
paint surface crossing the sky along the horizon. This ridge demarcates72 the limit
of a white layer of the sky skirting the large masses of vegetation and architecture73
but sometimes also passes beneath trees and the tops of certain painted buildings
(fi g. 2.34). The latter observation would not be surprising, were it not that this
white layer masks other buildings that were already painted during the fi rst
stage. Details of this painted architecture are visible in the ma-xrf maps and in
abrasions or micro-losses (fi g. 2.35). The white layer, which is at the same level
as the third layer of azurite higher up in the sky, covers them close to the horizon
Fig. 2.32a–d. Knights
(fi g. 2.45). We can therefore deduce that the third azurite layer belongs to the
of Christ: irr detail
second stage. The intermediate – and hence second – azurite layer is situated (a); vis detail (b);
between the fi rst stage and the second stage and might belong to either level. We visualization of the
think, however, that it passes beneath the fi rst-stage architecture and hence that it grey paint layer in a
belongs to the initial project, even in the absence of a sample to confi rm this. lacuna corresponding
On the other hand, a certain inconsistency has been introduced between the with the crags of the
wings and the central panel by the supplementary ultramarine and lead-white- fi rst stage (yellow
based layer, applied uniformly across the central panel and circumscribing the arrow (c); cross-
architecture (fi g. 2.36) or even overlapping details already painted (fi g. 2.37). The section 09/19 (white
presence of this fi fth blue layer raises questions: it breaks the greenish blue tone arrow) showing the
provided by the azurite that dominates the skies in the shutters. The question layer build-up as
of the ultramarine layer’s originality provoked discussion during the conservation follows: ground,
underdrawing and
campaign and has yet to be settled.74 It remains a delicate matter, moreover, because
priming (1, 2, 3);
the dove appears to belong to this stratum, even if the situation might be more crags of the fi rst
complex, as we will see in a moment. It should be added that this ultramarine layer stage (4); second-stage
was located beneath the overpaint in the sky that was removed during the 20016– meadow (5, 6, 7);
1 9 restoration campaign and which dated in any case to before 1 557.75 This mid- third-stage meadow
sixteenth-century overpaint, lastly, not only covered the ultramarine-based sky (8, 9); non-original
but also fi lled losses that were already present in this ultramarine-based stratum.76 varnishes (10–14) (d).
62
2. THE VAN EYCKS’ CREATIVE PROCESS
2.32a
2.32b 2.32c
14
13
10
0
12
11
6
5
4
3 2
50
0 μm
2.32d
63
2. THE VAN EYCKS’ CREATIVE PROCESS
2.33a 2.33b
Fig. 2.33a–b. Knights of Christ, detail: two painted birds
hidden by the meadow (probably third stage).
2.34a
2.34b 2.34c
Fig. 2.34a–c. Adoration of the Lamb. A ridge in the paint
surface along the horizon crosses some of the hills and
buildings. The diagrams show where it is visible in
raking light (red line) (a–b); detail vis (c).
64
2. THE VAN EYCKS’ CREATIVE PROCESS
2.35a 2.35b
2.35c
These losses are angular in shape, suggesting that the ultramarine layer mixed
with lead white had time to dry, become hard and crack. Even though a pictorial
layer rich in lead white polymerizes relatively quickly, the process still takes several
years – or even decades. We can therefore deduce that a certain amount of time
elapsed between the execution of the ultramarine sky and that of the sixteenth-
century overpaint. It may thus be assumed that the application of this ultramarine
sky, if not Eyckian,77 dates from the second half of the fi fteenth century, or the very
beginning of the sixteenth century at the latest.
Architecture
The superimposition of the two fi rst stages can be seen in the architecture on
the horizon of the central panel and the Knights of Christ panel. Coremans had
65
2. THE VAN EYCKS’ CREATIVE PROCESS
2.36a
already identifi ed Eyckian architectural elements and additions.78 His analysis was
partly fl awed, as he was unknowingly comparing the pictorial quality of buildings
overpainted in the sixteenth century (the tower of Utrecht cathedral) with others
that did not display the same overpainting (most notably the group of buildings on
the far left, referred to as ‘Coremans group 1’,, fi gs. 2.38, 2.39 and cve-1).79 However,
even after the recent removal of overpaints from these zones, it is undeniable that
different levels of intervention overlap and are juxtaposed on the horizon.
Based on the X-radiographs, Coremans believed that only for the two groups in
the central panel with large buildings (groups 1 and 7, and possibly also the lower
part of group 6; fi g. 2.38) were reserves left while the fi rst layers of sky and the
distant blue mountains were painted,80 and hence that they were planned from the
outset (cve 1 and cve 10). The new infrared refl ectogram and ma-xrf maps seem
to indicate that groups 5 and 6 were also planned from a very early stage, since a
fi rst painted version exists below the buildings now visible.81
First stage The large main building in group 1 already existed, surrounded by
houses, but with different architectural details – different windows and smaller
towers, for example. The upper part also underwent subsequent changes.82 Some
roofs, as well as a stepped gable to the right of the main building, had been
66
2. THE VAN EYCKS’ CREATIVE PROCESS
2.37a
2.37b 2.37c
Fig. 2.37a–c. Adoration of the Lamb, details (a–b) (Coremans group 1; see fi g.
2.38): the ultramarine-containing sky overlaps fi ne details of the architecture
(white arrows), among which a highlight on a spire; microphotograph (c).
2.38
Fig. 2.38. Adoration of the Lamb: Coremans’s diagram
numbering groups of buildings.
67
2. THE VAN EYCKS’ CREATIVE PROCESS
2.39a 2.39b
2.39c 2.39d
2.39e 2.39f
68
2. THE VAN EYCKS’ CREATIVE PROCESS
Fig. 2.39a–f. executed at this stage in red paint containing vermilion (fi g. 2.39f), others in
Adoration of the azurite (fi g. 2.39e). Similarly, the ma-xrf copper and lead maps, as well as the
Lamb: differences X-radiographs (cve-1), show that some of the mountains were initially painted
in group 1 between with a mixture of azurite and lead white (fi g. 2.39c and e), with space reserved
the fi rst and second
for the fi rst-stage roofs.83 The fi rst-stage buildings in groups 4, 5 and 6 can also
stages – vis (a);
fi rst stage visible
be seen through the small losses in the roofs of the houses painted later (fi g. 2.35)
in irr (b); ma-xrf as well as through certain buildings that have become more transparent over time
Pb-L map, some of (cve-11). We can also see via abrasions and losses that the initial windows, visible
the mountains were in the infrared macrophotograph and refl ectogram, were actually painted on top of
initially painted a fi rst paint layer, confi rming that they are generally not part of the underdrawing
leaving a reserve for (fi g. 2.400).
the fi rst-stage roofs The central area of the horizon, which includes the tower of Utrecht cathedral
(red arrows), and the and the new buildings discovered during the recent restoration campaign (fi g.
sky, too, was painted 1.21) was only occupied by hills. Observation of the X-radiographs, but especially
leaving a reserve the ma-xrf maps, now enables us to see that the Utrecht tower was indeed painted
for the fi rst-stage over a landscape with wooded hills that had already been painted to an advanced
architecture (c);
degree or even fi nished, given that we can make out painted trees wich were
vis diagram with
fi rst-stage visible
covered by the tower (fi g. 2.41). Part of the light blue hill from the initial concept
in irr (d); roofs can still be seen to the left of the tower. It is clear that in the fi rst stage there was
of the fi rst stage, less architecture but more wooded hills in the central part of the horizon of the
painted in azurite- Adoration of the Lamb.
containing blue The buildings making up group 7, lastly, also had a somewhat different appea-
visible in ma-xrf rance at fi rst, although the large volumes more or less correspond with what we
Cu-K map (blue see today (cve-10). Some conical roofs were painted blue-grey (fi g. 2.42) and the
arrows) (e); roofs colour of these fi rst roofs, based on azurite, is also visible in small losses in the
and stepped gable of upper layers. The same roofs also appear in the infrared refl ectogram, as do a whole
the fi rst-stage (red series of windows that differ from the ones we see now (cve-11).
arrows), painted A similar reworking of the architecture is found in the Knights of Christ panel.
in vermilion-
A reserve in the sky marks the contours of a fi rst stage comprising already painted
containing red
visible in ma-xrf
architecture, smaller in size and topped with two pointed roofs with yellow
Hg-L map (f). highlights (fi g. 2.43). One of the fi rst-stage roofs was not reworked and can still be
seen, protruding beyond the crag.
Second stage The horizon in the central panel was substantially altered following
the fi rst painted stage, which was largely completed. The second stage mostly
corresponds with what we see now, although certain elements were later reworked
once again.
The tower of Utrecht cathedral, the newly discovered buildings in the opening
of the landscape above the lamb, as well as the buildings in groups 1, 4, 5 and
6 are painted with a profusion of details that seem to us characteristic of Jan
van Eyck, rendered with deft touches, distributed coherently, never superfl uous,
pointing towards studies in situ of the architecture and even an understanding of
construction methods. It is most likely Ghent’s St Bavo’s Abbey that is depicted,
with well-integrated corner towers (cve-8).84 In the newly discovered ‘city’ (groups
5 and 6), the smallest detail is represented with great economy of brushwork. Two
touches were all that was needed, for instance, to represent the entire volume of
69
2. THE VAN EYCKS’ CREATIVE PROCESS
2.400a 2.400b
2.400c 2.400d
Fig. 2.400a–d. Adoration of the Lamb,
detail (a and c) (Coremans group 5;
see fi g. 2.38); windows visible in ir (b)
(arrows); these fi rst-stage windows
were painted on top of an ochre-pink
paint layer (arrows) and covered by the
second-stage architecture (d).
70
2. THE VAN EYCKS’ CREATIVE PROCESS
Fig. 2.41a–c.
Adoration of the
Lamb: hills and
trees, including
cypresses, of the fi rst
stage, hidden by the
Utrecht tower and
still visible today –
ma-xrf Cu-K map
(a); ma-xrf Pb-L
map (b); vis (c).
2.41a
2.41b
2.41c
71
2. THE VAN EYCKS’ CREATIVE PROCESS
2.42a 2.42b
Fig. 2.42a–b.
Adoration of the
a dormer window emerging from a roof (fi g. 2.44). The judicious placement of Lamb, detail (a);
highlights gives shape to architectural details and helps distinguish a wide variety conical roofs
of materials – stone, brick and roofs covered with glazed tiles or copper. It is not painted blue-grey
impossible, however, that Jan retained certain elements of the architecture from in the fi rst stage,
the initial project and incorporated them into his alterations. now hidden but
These second-stage buildings were sometimes painted directly over those of the visible in the ma-
fi rst stage or onto a light intermediate layer corresponding, as we have seen, with xrf Cu-K map (b).
the fi nal azurite-based layer of the sky (fi g. 2.45). This overlap led to differences
in material density between the lower parts of the buildings, painted on a dark
underlayer corresponding to the mountains or buildings of the fi rst stage, and the
upper parts, painted over this light, second-stage sky. This is the case for the tops
of the main building and the bell tower to the left of the large fi g tree in Coremans
group 1, for the tower of Utrecht cathedral, and for the majority of the architecture
in the right half of the composition.
Third stage Other changes have been made, which cover the second-stage painting,
mostly by making it heavier in appearance. They include windows accentuated
by touches of opaque dark paint and roofs repainted without nuance. In the large
building of group 7 in particular, the pointed roofs stick out from the rest (fi g.
2.42a). These are painted in a fairly opaque red paint and seem stuck on rather
than truly attached to the body of the building, as we mentioned in the case of
St Bavo’s Abbey. The black lines representing the openings in the walls of this
weaker architecture are repetitive shapes, far removed from the deft touches found
elsewhere. This intervention appears to have been done by a painter who only
reworked this zone locally. The hexagonal tower and round building, for example,
72
2. THE VAN EYCKS’ CREATIVE PROCESS
Fig. 2.43a–e.
Knights of Christ:
fi rst-stage building
covered by second-
stage architecture
– vis (a); irr (b);
diagram with
details visible in
irr (c); rx diagram
with details visible
in irr (d); vis
diagram with
details visible in
irr (e).
2.43d 2.43e
73
2. THE VAN EYCKS’ CREATIVE PROCESS
2.44a 2.44b
Fig. 2.44a–b.
Adoration of the
are more refi ned in quality, even if a reworking on the left side of the domed roof Lamb, detail: two
betrays the intervention of a weaker hand (cve-12). touches of paint
This painter, who strikes us as less skilful, tried to reproduce what he observed suffi ce to represent
in the arrangement of the original architecture. And when he did make noteworthy the volume of
changes, he sought to integrate them into the ensemble already painted by adding dormer windows.
numerous, scattered touches, not only in the architecture on the horizon but also in
the vegetation where the landscape intersects with the sky. In this way, architectural
elements from three stages seem to coexist on the horizon of the central panel.
The city in the distance of the central panel (group 6) belongs to the stage of Fig. 2.45. Diagram
the ultramarine sky and is itself painted with ultramarine. It was done after the of the paint layers
stage-1 and 2 buildings had already been painted (fi g. 2.46). A similar reworking of the sky and
of the city in the distance in the Pilgrims (cve-29), painted between a summarily concordance with
two levels of painted
architecture.
Azurite 3
Azurite 1
Indigo
2.45
74
2. THE VAN EYCKS’ CREATIVE PROCESS
2.46a
2.46b 2.46c
Fig. 2.46a–c.
Adoration of the Lamb:
the ultramarine layer executed dark blue foreground and mountains in the distance,85 covers a fi rst
(sky and architecture landscape86 done in lighter and warmer tints and can be made out between the
of distant city) brushstrokes of the urban landscape we see today. Some buildings in the distance
skirts and slightly seem to have been planned from the underdrawing stage: the infrared refl ectogram
overlaps the painted of this area appears to confi rm this (cve-29), unless what we see is – as elsewhere –
architecture. the painted realization of a fi rst city. The silhouettes of the buildings are summarily
rendered and the light touches to suggest stretches of wall or roof lit by the sun are
placed randomly, without introducing volume or perspective, contrary to what is
seen in the majority of the buildings in the central panel. Similarly, the dark-blue
foreground is painted without nuance. It reprises the idea of the fi rst level in the
stratigraphy of the underlying landscape painted in green, but without introducing
variations. However, the dark-blue stripe and the architecture pass below the trees
placed in front, whether it be the cypress, the palm or the small trees closer to
the city skyline. The painter who executed this city retouched and enlarged the
vegetation painted in front of it to further integrate his addition. The infrared
macrophotograph and the refl ectogram show a difference in the distribution of
75
2. THE VAN EYCKS’ CREATIVE PROCESS
black pigments in the branching of the cypress in front of the city, suggesting that Fig. 2.47a–b.
these were reworked (cve-29). Hermits: fi rst-stage
The horizon of the Knights of Christ was then also modifi ed after a fi rst painting tree trunks in the
that might have been completed. The blue mountains and dark-brown vegetation ma-xrf Pb-L map (a)
covered with second-
are similarly painted with few nuances: we will return to them in chapter 4.
stage green foliage
– vis (b).
Trees
First and second stages Some adjustments to the landscapes were already noted
by Coremans and Van Asperen de Boer, particularly in the thickets of trees.87 The
infrared images show a larger number of trunks than we see today and their volume
already seems to have been well established, as the position of knots in the wood
(cve-2), for instance, was already indicated.88 In some cases the trees had even been
painted, as in the Hermits and Pilgrims, where a certain modelling is visible in
the X-radiograph and the ma-xrf maps (Ca-K/Pb-L/Fe-K/Zn-K) (fi g. 2.47, cve-
24 and cve-27), implying a fi rst pictorial layer at least. These more numerous
fi rst-stage trunks were then partly concealed by foliage in the shutters or, in the
Adoration, completely masked by fl ower beds painted on the second-stage meadow.
To give an example, a small tree planned for the fi rst stage in the central panel was
ultimately replaced by a herbaceous plant (cve-2).89 The mass of foliage to the left
of the confessors in the central panel, which now consists of a large fi g tree, rose
bushes and pomegranate trees, must have looked more like a thicket of trees with
the base of the trunks visible, similar to the existing ash and almond trees90 to
the right of the same group. The infrared refl ectogram reveals what this fi rst idea
looked like, with visible trunks (cve-2).91 The infrared refl ectogram made during
the treatment also shows the presence of two large trees – likewise with visible
trunks – beneath the tower of Utrecht cathedral (fi g. 2.48). They are now hidden
by the smaller bushes of the second-stage meadow. It is hard to make out whether
these trees were also painted: the ma-xrf maps do not confi rm this and the zone in
question is also one of the most damaged parts of the panel.
All these trees with visible trunks, which were intended to provide a certain
continuity between the central panel and the shutters, appear to belong to the fi rst
stage, given that, in the central panel, they were covered by the second meadow
and the fl owers dotting it. It is hard to determine the extent to which these trees
with visible trunks were painted or not.
Coremans also noted that some trees92 in the landscape on the right of the Knights
of Christ conceal an initial landscape organized in three receding planes.93 This
landscape was already painted, since it is visible in the X-radiograph and the lead, Fig. 2.48a–b.
Adoration of the
copper and iron ma-xrf maps, in which a path can be made out meandering across
Lamb, detail (a);
this preliminary landscape (cve-200). A sample taken from an area of meadow two fi rst-stage trees
shows a superimposition of six green paint layers, including a glaze above three can be made out
yellow-green layers. The glaze has itself been covered with an opaque layer and between the tower
then again with a green glaze, as seen in the meadow of the central panel, which of Utrecht cathedral
tends to confi rm that the fi rst landscape had already been painted to an advanced and the altar in the
stage or even completed. irr (b); see also cve.
76
2. THE VAN EYCKS’ CREATIVE PROCESS
2.47a 2.47b
2.48a 2.48b
77
2. THE VAN EYCKS’ CREATIVE PROCESS
2.49a 2.49b
Fig. 2.49a–b.
Pilgrims: plain
Zones with thickets of trees requiring a dark base were constructed using a black areas serve
similar process in the central panel and wings. First of all, plain areas of black as dark bases for
were brushed on quickly.94 These are clearly visible in the infrared refl ectograms thickets of trees –
of the Pilgrims and Hermits (fi gs. 2.49 and 2.500). The limits of this rather thick vis (a); irr (b).
pictorial material can even be made out by the naked eye through the pictorial
layers (cve-26) or where it has not been covered along the edges of the paintings
(fi g. 2.500). Uniform tones like this below the thickets or patches of vegetation do
not stand out as clearly in the infrared refl ectogram of the central panel, but they
were nonetheless undeniably applied in order to position zones with thickets of
trees on either side of the groups of confessors and virgin martyrs. The difference
in rendering probably relates to the typology of the wooded zones: less dark for
thickets and darker for deep forest.
The application of large fl at, monochrome areas in preliminary layers is a common
practice in Jan van Eyck.95 The technique seems to have been applied before Jan,
possibly even by Hubert, as the black base of the spring would appear to suggest.
Zones of dense vegetation in which an area of fl at black was laid down vigorously
before the green foliage was painted can be found, for example, in the St Christopher
in the Antwerp–Baltimore Quadriptych, done around 13800–1400 00 (fi g. 2.51).96
The foliage of the trees was then painted over these fl at black areas, while the
leaves that stand out against the sky were not painted directly onto the light layer
of the sky but over an intermediate translucent brown layer (fi g. 2.52). Strictly
speaking, these are not rapidly brushed fl at areas like the black base tones, but
brown dabs97 of varying thickness, which already correspond more or less with the
78
2. THE VAN EYCKS’ CREATIVE PROCESS
2.500a
2.500b 2.500c
Fig. 2.500a–c.
Hermits: plain black
shape of the foliage to be added in green on top. An intermediate brown layer of
areas left uncovered
along the borders
this kind also seems to have been used to create the leaves of the cherry tree visible
of the panels – in the window of the Arnolfini Portrait98 or in the vegetation of the Virgin at the
vis (a); irr (b); Fountain.99
detail vis (c). The trees in the Adoration of the Lamb, especially the cypresses and the leaves
that stand out against the skies, are painted in green over those brown layers,
which ensure the transition between the dark masses of the trees and the light
area of the sky by introducing a kind of aerial perspective. ma-xrf maps show
the presence of copper, iron and zinc in these brown glazes. These elements were
systematically found in all the brown layers used as transition between masses of
vegetation and the sky (fi gs. 2.52 and 2.53). Applying the green foliage directly,
without such a transition, over the areas of fl at black and over the sky, would have
79
2. THE VAN EYCKS’ CREATIVE PROCESS
2.51a 2.51b
Fig. 2.51a–c.
Antwerp–Baltimore
Quadriptych:
St Christopher
(Antwerp, Museum
Mayer van den
Bergh) – vis (a);
detail vis (b); irr (c).
2.51c
resulted in overly strong contrasts that would probably have disturbed any formal
continuity. A similar practice is already found in pre-Eyckian works, such as the
Tower retable with Scenes from the Infancy of Christ in the Museum Mayer van den
Bergh, Antwerp.100
The fl at areas of black were laid down after completing at least one layer of the
sky.101 The brown layer of the vegetation was painted over the fi nal azurite layer. The
lead-white-rich layers constituting the sky were applied with wide brushstrokes
that imprinted streaks in the pictorial material. This striated effect is accentuated
by the fl uid brown glaze, which settled more thickly in the hollows of the striations
(cve-26).102 Use has been made of the black underlayer itself in the fi nal appearance
to convey the density of the wood: it serves as a base for the yellow-green leaves
painted with rapid touches (cve-25). The fl at areas of black in the central panel are
also exploited using a sgraffi to technique, in which the green paint layer of foliage
overlying areas of unmodulated black was scraped with a stylus or the back of the
brush while fresh to expose the underlayer. The stems of the pomegranate tree to
the left of the confessors were rendered this way before the leaves themselves were
painted (fi g. 2.54).
All these technical characteristics, the application of large areas of unmodulated
colour in the fi rst layers, use of a glaze over a striated underlayer and the sgraffi to
technique applied in the fresh paint layer, are found in Jan’s repertoire,103 although
they may also have been practised by his brother. They exist in pre-Eyckian
paintings, too, for that matter.104 Considering that the fi nal layer of sky in the
shutters belongs to the second stage, however, the trees painted on top of the sky
80
2. THE VAN EYCKS’ CREATIVE PROCESS
2.52a
2.52b 2.52c
Fig. 2.52a–c.
Pilgrims detail (a):
brown glazes must be contemporary with or later than that stage. A sample taken from the
producing a strong small palm tree on the left in the Pilgrims shows that it is painted on a sky already
zinc signal comprising the four layers found in the skies of the wings. On the other hand, the
(ma-xrf Zn-K) (b); fi rst paint layers might belong to the fi rst stage.
brown glazes (white Coremans noted, based on X-radiographs, that during the execution of the sky no
arrows) vis (c). space was reserved for some trees of Mediterranean origin – namely the palm tree and
the four relatively large cypresses on the horizon of the central panel. The umbrella
pines, palm trees, cypresses and orange trees in the Pilgrims and Hermits were not
originally planned either.105 Coremans therefore deemed them to be additions
made by Jan van Eyck after his time in Spain (i.e. from 14300 onwards) to a more
northern landscape attributable to Hubert. Jan van Eyck is known to have travelled
to the Iberian Peninsula in 1428–29 (with the embassy sent by Philip the Good to
Portugal in order to fi nd a new bride for the duke), and perhaps already in 1426.106
He spent at least ten months there and had ample time to familiarize himself with
and sketch the gardens and vegetation of the Mediterranean region.107 We have just
81
2. THE VAN EYCKS’ CREATIVE PROCESS
2.54a 2.54b
82
2. THE VAN EYCKS’ CREATIVE PROCESS
seen that the smallest of the palms in the Pilgrims panel was indeed painted late
into the execution. The large central palm tree in this panel was itself added over
the already painted hills: its trunk partially covers the overlapping branches of the
thicket of trees on the right. The rendering of the typically rhomboid pattern of a
date palm trunk, in both the large palm and – albeit more schematically – the small
palm on the left, suggests a good knowledge of this species, which did not grow in
northern Europe in the fi fteenth century.108
These areas of vegetation in the central panel and wings were therefore altered
at an already fairly advanced stage of painting. The way the thickets of trees
were represented in the fi rst stage seems in keeping with a more conventional
conception of the medieval landscape, with rows of straight, woody trunks. The
changes introduced highly varied vegetation, possibly represented less repetitively
and hence with greater animation. They are particularly noticeable in the central
panel, with the remarkably rendered fl owerbeds and the addition of essentially
Mediterranean trees, although it is important to note that cypresses were already
painted on the fi rst-stage hills before the tower of Utrecht cathedral was added
(fi g. 2.41). All these modifi cations are contemporary with the second meadow and
the stone fountain, since the fl owers that conceal the old trunks of the thickets
of trees are contemporary with the second-stage meadow. It is more diffi cult, by
contrast, to comment on the progress of the trees before Jan resumed the work and
to determine the extent to which he was able to realize most of them or simply
to insert fi nishing touches onto plain black areas that might already have been
painted. The lack of comparative elements in Jan’s work, and a fortiori in that of
Hubert, makes the task even less straightforward.
Third stage A large amount of reworking that in our view is alien to Jan’s manner
is found across a substantial part of the vegetation.109 These touches, done using a
fairly opaque and grainy material,110 were rather systematically applied to the surface
of vegetation already painted. They consist merely of repetitive touches without
any real connection to the foliage they are supposed to represent. To the right of the
large palm tree in the thickets in the Pilgrims (fi g. 2.55), these imprecise touches
fi ll the black spaces of the underlayer and introduce separations within this mass of
trees. Comprising a light brown-green paint, they produce a strong tin (Sn) signal
in the ma-xrf maps.111 The same desire to fi ll in the spaces left without foliage
resulted in the systematic addition of similar touches in the Hermits (fi g. 2.56), and
they are also present in the bushes on either side of the path descending from the
thicket in the Pilgrims, where they are superimposed on leaves that seem to have
been built up mainly using green glazes. These touches are likewise found in many
of the cypresses (fi g. 2.57) and in the trunk of the large palm in the Pilgrims, as well
as in the Knights of Christ in certain trees painted over the landscape, which initially
featured a painted path.112 The latter trees, too, appear to have been painted at fi rst
with green glazes and fewer highlights, and then covered with the same touches,
which produce a strong tin signal in the ma-xrf map.
The central panel is not free from such stippled (fi g. 2.58) or even more signifi cant
reworkings, as in the palm tree, the much more summary rendering of which113
lacks the realistic features of the trunk of the large palm in the Pilgrims.
83
2. THE VAN EYCKS’ CREATIVE PROCESS
2.55a 2.55b
Fig. 2.55a–b.
Pilgrims, detail:
third-stage touches.
84
2. THE VAN EYCKS’ CREATIVE PROCESS
First and second stages The majority of decorative effects are found in the area of
the central panel with the large fi gures and chiefl y comprise the rendering of the
fabrics worn by the philosophers, the patterns and textures of which were achieved
85
2. THE VAN EYCKS’ CREATIVE PROCESS
2.58a 2.58b
Fig. 2.58a–b. Adoration of
the Lamb, detail: third-stage
touches.
2.59a 2.59b
Fig. 2.59a–b. Adoration of the
Lamb, detail: diagram of third-
stage touches on top of a partly
hidden earlier stage with brown
glazes outlined in white (a);
brown glazes detected in the
ma-xrf Zn-K map, during
restoration (b).
86
2. THE VAN EYCKS’ CREATIVE PROCESS
2.600a 2.600b
Fig. 2.600a–b.
Adoration of the
Lamb, detail (a);
decorative patterns by working wet-in-wet. There is a multiplicity of shapes and pictorial effects,
realized by scoring from decorative motifs scored into the fresh paint (fi g. 2.600), a pattern of tiny
the fresh glaze dots created with a stick or the back of the brush into the fresh glaze (fi g. 2.61) to
with the back beading effects of the paint layer (fi g. 2.62b). The lines or dots scored into the fresh
of the brush in paint to reveal the underlying paint layer are widely used in the large fi gures in
order to reveal the the foreground to the left of the fountain. They were added in the fresh glazes to
underlying paint create patterns in the hats (fi g. 2.62c) or to simulate the texture of the fabric in the
layer (b). red robe of the large fi gure in the lower left corner (fi g. 2.61).115 A similar practice,
less frequent in the other parts of the lower register of the open altarpiece, was also
detected in some places in the panels of the exterior, notably in the pages of the
prophet Micah’s book.116 The process is specifi c to oil painting, because only an oily
binder, which takes a certain time to dry (unlike all other painting systems, where
the diluent is aqueous and drying is instant), offers the painter this possibility. The
very fact of its use indicates a certain dexterity of handling.117
The use of additives118 to accelerate the drying of the paint and the presence of
drying crackle in certain fi elds of colour of the Ghent Altarpiece show that the Van
Eyck brothers were fully aware of the technical problems intrinsic to oil painting.119
What is more, we can safely assume that they were familiar with the demanding
rules of the medium and the fl aws that may result from not observing them. They
actually used these fl aws to good effect at times, including the beading effects that
occur when a fresh layer is applied to a pictorial layer that is already dry on the
surface or when the fat-on-lean rule is not respected.120 We observed this in certain
87
2. THE VAN EYCKS’ CREATIVE PROCESS
2.61a 2.61b
88
2. THE VAN EYCKS’ CREATIVE PROCESS
2.62a 2.62b
2.62c
89
2. THE VAN EYCKS’ CREATIVE PROCESS
Third-stage changes Modifi cations are found in the large fi gures in the foreground
of the central panel, particularly where the kneeling fi gures meet the standing
ones. To the left of the fountain, the fi rst standing fi gures with green and pink
draperies were emphasized to the detriment of three kneeling silhouettes, which
had already been painted, perhaps even completely (cve-4). This is slightly visible
to the naked eye and more in the technical documents.123 The kneeling fi gure in
the blue cap wore a robe that was tied around the waist with a large belt, similar to
that of the fi rst fi gure to the immediate left of the fountain. The green drapery was
added, concealing much of the original fi gure. A similar adjustment was made to
the pink robe and chaperon of the fi gure above to partly obscure the heads of two
90
2. THE VAN EYCKS’ CREATIVE PROCESS
kneeling men. The standing fi gures dressed in green and pink were thus enlarged
to the detriment of these three kneeling fi gures, which are now barely visible. The
effect is to create a clearer separation between the kneeling and standing men. A
similar change can be seen in the group to the right of the fountain, where the red
drapery of the pope standing behind the kneeling apostles has been rigidly plastered
over the back of one of the kneeling fi gures (cve-14). Were these changes made
simply to adjust the articulation of the foreground or to mask possible damage?
The question can be posed in relation to a zone near the left hand of ‘Virgil’ (clad in
white, with laurel wreath), in which there is a lacuna embedded in the underlying
paint layers (fi g. 2.63).124
Coarse touches have been added, lastly, to some of the fi gures in the foreground
of the central panel at the interface between the meadow and the fi gures, probably
to ‘improve’ the juxtaposition of the fi rst and second stages. They occur particularly
at the point where the second-stage meadow skirts the large fi gures, thickening the
pictorial layer. Some touches add stiff hair (fi g. 2.64) or heavy highlights to the top
of certain skulls (fi g. 2.65).
The fl esh tones of St Christopher in the Pilgrims panel appear to have been reworked.
In an already painted fi rst stage, the eyes were positioned differently, the saint
looked over his shoulder and his nostrils were more visible (cve-28 and ma-xrf
Pb-L). The giant’s hair was extended over his right shoulder, concealing the top
of a fi gure’s head behind him. The obscured head shows a strong signal of iron in
the ma-xrf map as it was painted brown – probably like the hair of the young boy
in the red cloak holding a stick in the crowd of pilgrims. Eyckian features such as
Fig. 2.64a–b. highlights emerging from the shadows125 can be found at the base of St Christopher’s
Adoration of the nose, while other, coarser touches belonging to the third stage have been added to
Lamb, detail: third-
stage retouchings.
2.64a 2.64b
91
2. THE VAN EYCKS’ CREATIVE PROCESS
2.65a 2.65b
2.65c 2.65d
Fig. 2.65a–d.
Adoration of the
his white turban. The latter resemble the enlargement of some shapes in the body Lamb, detail: heavy
of the lamb and on the altar, which we will mention in a moment. highlights applied
Signifi cant elements were modifi ed in the Knights of Christ panel in the group of to some of the
four fi gures on the left at the foot of the large rock and also in the legs and hooves, heads.
eyes, ears, muzzles, manes and gear of the horses and in the standards carried by
the knights.126 The ma-xrf maps and X-radiograph confi rm that these adjustments
were made primarily during the painting process. We have already mentioned the
Fig. 2.66a–b.
initial position of the horses’ hooves in the fi rst stage, in which the ground differed Knights of Christ,
from the current brown soil.127 These changes appear to have been made in the third-stage purple
second stage to horses painted in the fi rst stage.128 cloak (a) painted
Certain other changes, by contrast, belong to the third stage. This is particularly over a brocade
the case for the knight with the bushy grey beard, whose purple, ermine-trimmed garment visible in
cloak was painted over a brocade garment, the patterns of which were painted the ma-xrf Fe-K
and show up in the ma-xrf map of iron and zinc (fi g. 2.66).129 A cross-section map (b).
92
2.66a 2.66b
2. THE VAN EYCKS’ CREATIVE PROCESS
reveals the stratigraphy of the corresponding pictorial layers.130 The brocade was
overpainted with the purple garment we now see, which alters the long green
sleeve of the harnessed knight, but does not seem entirely to cover the bottom part
of the original brocade robe (cve-19 and fi g. 4.18). It may appear surprising that
the brocade pattern would have been overpainted with a rather uniform layer like
this,131 lacking any decorative pattern. The intention may have been to create a less
elaborate background, so that the muzzle of the horse painted down the left edge
would stand out more. This muzzle was meant as a continuation of the partially
represented head of the second white horse on the right of the Just Judges (fi g. 4.17).
Painted over the purple robe, its anatomy is noticeably less well observed than
those of the other horses’ heads.132
Other layers added to fi gures in this zone133 raise similar questions, including the
shapeless retouching that masks part of the crown of the knight in green brocade
(cve-17);134 the bright red robe of the fi gure upper left (cve-18), which was painted
over a deep-red glaze; the hair of the same fi gure, which covers an already painted
ear and a red cap worn below the crown; and, perhaps, also the ultramarine chaperon
of the fi gure referred to as ‘Six’,135 which covers an azurite chaperon adorned with a
crown (fi g. 3.9).136
The Lamb
The sixteenth-century overpainting137 of the lamb was partially taken away in the
1 950
0s to reveal the original ears and has now been completely removed.138 This
allowed the recovery of the animal’s eyes with horizontal slit pupils typical of prey
animals (fi g. 1.8c).139 The body of the lamb was painted in a thin white layer with
brown lines to suggest separate tufts of hair, followed by scattered white touches to
evoke woolly curls, all this together evoking the thickness of the fl eece.
The recent restoration campaign also revealed that the lamb’s body has been
enlarged compared to its original, less solid form. This can be made out in raking
light (fi g. 2.67) and in the infrared images and ma-xrf maps, especially the copper
map (fi g. 2.68), and was recently presented in a study combining multispectral
images and ma-xrf maps.140 The enlargement of the lamb’s body actually overlaps
the surrounding second-stage meadow (fi gs. 2.67 and 2.68). The green layer of
this second meadow seems to have skirted carefully around the original form of the
lamb, creating a ridge along that form, as is also the case around the large fi gures
of the apostles and prophets. The animal’s body was enlarged using a thick white
layer, especially in its hindquarters, along the upper edge of the spine and the top
of the head (fi gs. 2.69 and 2.700). These additions were made with brushstrokes
that suggest tufts of white hair at the back, but without distinguishing the hairs
as elsewhere in the body. The touches in question recall those we already noted
on top of the apostles’ and philosophers’ skulls (fi g. 2.65). There is also a brown,
translucent shadow enlarging the left side of the body. The enlargement of the
lamb’s body thus appears to belong to the third stage. The shape visible in the
infrared refl ectogram follows probably more closely the original shape, which
seems more realistic (fi gs. 2.700 and 2.71).
94
2. THE VAN EYCKS’ CREATIVE PROCESS
2.67
Fig. 2.67. Adoration
of the Lamb, detail:
enlarged body of the
lamb overlapping The lamb’s gilded rays are located above the additions in the animal’s head and
the second-stage along its neck and hence also belong to the third stage. Observation of this zone
meadow noticeable in raking light reveals a demarcation around the head of an irregular halo-shaped
in raking light (after form extended in height (fi g. 2.72). This irregular halo is also visible in the ma-xrf
treatment).
lead and tin maps, as well as in the X-radiograph. Perhaps it indicates a change
to mask something that was originally planned or else it may be the trace of an
intervention to address damage caused by the presence here of a dowel pin in the
support.141 None of the currently available scientifi c imaging techniques – not even
ma-xrf – enables us to determine whether a halo or other rays were worked out
in the fi rst or second stage. However, rays do appear to have been intended in
the underdrawing – the infrared refl ectogram allows some of them to be detected
(fi g. 2.73).
95
2. THE VAN EYCKS’ CREATIVE PROCESS
2.68
Fig. 2.68 Adoration
of the Lamb, ma-xrf
Cu-K map (during
restoration) showing
the enlargements
of the lamb’s body
overlapping the
second meadow.
Fig. 2.69 Adoration
of the Lamb,
detail: third-stage
enlargement of the
lamb’s body (white
arrows).
2.69
96
2. THE VAN EYCKS’ CREATIVE PROCESS
2.700a
2.700b
2.700c
97
2. THE VAN EYCKS’ CREATIVE PROCESS
2.71
Fig. 2.71.
Adoration of the
Another signifi cant change was observed in the altar,142 where a preliminary Lamb, detail:
version had once again been painted during the fi rst stage. This fi rst version of third-stage
the altar had a deeper top (cve-9) and the perspective was different. It was skirted enlargement of
by the second-stage meadow, creating a ridge in the paint layer visible in raking the lamb’s body
light. The perspective of the altar table was later adjusted, reducing its depth, by (white arrows)
seen in raking
the placement of a strip of green painted meadow, and the sides were adapted.
light.
These interventions seem to have occurred after the second-stage meadow had
been painted. All the indications are, therefore, that the position of the altar was
adjusted at the same time that the body of the lamb was enlarged.
The issue of the dove, the halo and the rays is a complex one: various stages
of elaboration can be detected, especially in the rays, even after removal of the
sixteenth-century overpaint, which included an additional level of gilded rays.
The fi rst-stage rays were established by the preparatory drawing (approximately
28 rays143) and then scored into the ground and isolating layer (fi g. 2.74a). Their
lengths vary144 so as to form a symmetrical star pattern (fi g. 2.74b). The drawn and
scored rays are not interrupted beneath the body of the dove, and the dove does
not show any preparatory drawing in the infrared refl ectogram (cve-6). These rays
were abandoned in the second stage.145 No trace of gilding, nor even of a mordant,
98
2. THE VAN EYCKS’ CREATIVE PROCESS
2.72a 2.72b
Fig. 2.72. Adoration of the Lamb, detail: damaged area around the head of the lamb
seen in raking light (a) and in the ma-xrf Pb-L map (during restoration) (b).
2.73a 2.73b
Fig. 2.73. Adoration of the Lamb: presence of rays at
underdrawing stage (white arrows) – vis (a); irr (b).
99
2. THE VAN EYCKS’ CREATIVE PROCESS
2.74a
2.74b
was detected on top of any of the three azurite-based paint layers in the sky of the Fig. 2.74a–b. Diagram
of rays scored into the
central panel.146 It was only after the ultramarine layer of the sky and halo had been
ground (green lines),
painted in the third stage that the gilded rays were fi nally executed. This set of visible in the xr (a);
gilded rays differs from the original drawing: they are much more numerous (129) the incised rays of
and the original star-shaped pattern has been abandoned. If the decision fi nally to variable length form
add gilded rays was non-Eyckian, the original idea might still have been known, as a symmetrical star
the reliefs impressed in the paint layer along the scored but not painted rays could pattern (schematically
perhaps be seen before the ultramarine layer was applied, just as we can still today illustrated with purple
distinguish some of them in raking light. lines) (b).
100
2. THE VAN EYCKS’ CREATIVE PROCESS
2.75a
2.75b 2.75c
Fig. 2.75a–c.
Comparison of the
dove in the Adoration
of the Lamb, detail (a);
It thus appears that the idea of providing the sky with gilded rays was abandoned
and in the Virgin in the second stage. On the other hand, no trace of clouds or birds was found at
Annunciate on the the level of the azurite-based layers of the central panel, in contrast with the skies
exterior, detail (b); of the wings, where they can be seen on the surface. Could the sky of the central
losses in the thick panel have been painted in the second stage without clouds, birds or rays? This
white top layer of hypothesis seems plausible when we consider that a halo, or even a dove, which
the dove in the were not apparently planned in the underdrawing, were added at this time: the
Adoration of the Lamb, halo we see now belongs to the same stratum as the ultramarine sky, and it is also
microphotograph (c). hard to ascribe the present dove to Jan van Eyck (fi gs. 2.75a, 2.75b and 4.19).147
Some scholars have raised the possibility that no dove was present in the original
concept.148 No one has suggested the presence of a preliminary dove or halo that
would have been subsequently concealed by the corresponding elements we see
today.
Samples taken from the dove indicate that it was painted in two layers and even
three in certain places.149 Small losses can be seen under the stereomicroscope in
the white pictorial layer applied thickly to certain parts of the dove (fi g. 2.75).
An underlying white layer is visible through these losses. Each colour of the
halo consists of two layers, both of which contain a mixture of ultramarine to
which other pigments have been added to create the different nuances of the halo
(fi g. 2.76). The two ultramarine-based layers are located on top of the four layers
of sky, the fi rst of which is based on indigo and the following three on azurite. One
of the questions that arises relates to the fi rst of the two ultramarine-containing
layers. Does it belong to the stratum of the ultramarine sky, which we consider to
be non-Eyckian? If so, this would make the dove non-Eyckian. Or does it belong
to an earlier stratum, applied locally by Jan in order to distinguish the area around
the dove, using some ultramarine, from the remainder of the sky based on azurite?
Careful observation of this important zone does indeed lead us to wonder whether
101
2. THE VAN EYCKS’ CREATIVE PROCESS
Dove
Halo
Azurite 3
Azurite 2
Azurite 1
Indigo
2.76
Fig. 2.76. Diagram
– Hypothetical
there is not a difference between the ultramarine of the sky and that used below build-up of the
the dove and inside the halo around the dove. The laboratory analysis does not, paint layers in the
however, allow us to conclude one way or another. halo and the dove.
Yet another question concerns the pictorial layers of the dove. There is no evidence
to state with any certainty that there is a dove below the present one.150 But could
the lower layer be Jan van Eyck's work? Analysis of the additives, and in particular
of the powdered glass found in many of the altarpiece’s paint layers, reveals the
presence of particularly large wood-ash glass particles in the upper layers (up to
900 µm), whereas no glass particles have been detected in the lower, sometimes more
greyish layer found in three of fi ve cross-sections made in 1951. It should also be
noted that in the only sample taken from the body of the lamb powdered glass was
not detected.151 The glass detected in the upper layer of the dove corresponds with
a type found very frequently in the pictorial layers of fi fteenth-century Flemish
paintings. As it differs somewath from the glass found in the second stage paint
samples,152 it might even postdate the Van Eycks. The fi rst layer of the dove could
belong to the second stage.
Conclusion
It is apparent after the recent treatment that three stages can be identifi ed in the
development of the lower tier of the altarpiece interior. The fi rst of these stages is
barely visible now, as it has been concealed by the subsequent interventions. In
addition to the construction of the supports and the completion of the preparatory
layers, a large proportion of the fi rst layers of paint date from this stage, not only
in the skies and the vegetation of the central panel and the wings, but also in the
positioning of the principal human and animal silhouettes. A different conception
was planned for the central axis of the Adoration panel, with a simple spring in
the ground and a sky without a dove but with rays, albeit few in number, all in a
102
2. THE VAN EYCKS’ CREATIVE PROCESS
less varied landscape. This stage appears to have remained unfi nished and only few
elements of it are now visible, given the extent to which they were covered by the
second-stage intervention.
The second stage followed much of the fi rst-stage composition, but introduced
signifi cant changes to the central panel by adding an imposing fountain above the
spring and possibly a dove and halo. It also abandoned the idea of rays in the sky.
Elsewhere, the main changes relate to the adjustment of the architecture on the
horizon, which is more numerous in this stage, the addition of more abundant
and varied vegetation, and the fi nalization of the human fi gures, with occasionally
signifi cant adjustments of form and colour.
Touches of varying size, lastly, were added to all the paintings, in order to
‘improve’ the perspective, particularly along the central axis of the Adoration panel,
to ensure continuity between the different shutters and undoubtedly also to satisfy
a certain horror vacui. All the indications are that these touches post-date the second
stage, over which they have been placed. These additions include the current sky of
the central panel, as well as the current dove and halo.
The three different stages observed here were distinguished because they are
superimposed in many places, but in some areas they are also juxtaposed.
Notes
1 Jones, Augustyniak, Dubois 200200. 11 Coremans 1953, pp. 12–15; Glatigny
See also the Introduction to this volume. et al. 20010, pp. 172–232.
2 The date of Hubert’s death is known 12 See chapter 3, section ‘Ground’.
from his epitaph. Duverger 1945, 13 Postec, Steyaert 200200, pp. 196–99;
pp. 15–28; Martens 20019, pp. 125–26; Postec 20012, pp. 148–49.
Paviot 200200, p. 600. 14 Some are visible to the naked eye, and
3 The Just Judges were stolen in 1934. they show up in the irrs as grey lines
4 Panofsky 1953, pp. 2007–22. Panofsky or in the xrs as black lines, due to the
refers to earlier authors in note 2008/3. absence of pictorial material.
5 Fraiture 20017. 15 Postec, Steyaert 200200, pp. 196–98.
16 It should be noted, though, that bristles
6 The present study has benefi ted from are fairly easy to identify in the irrs of
the comments and observations of all the altarpiece exterior but not in those
the conservators who took part in the of the paintings of the lower register of
treatment of the four paintings; see the the altarpiece interior, as the latter are
Introduction to this volume. far more detailed, making it harder to
7 Postec, Steyaert 200200. distinguish possible bristles.
8 Coremans 1953; Van Asperen de Boer 17 A technical detail not found in the panels
1979. Without seeking to be exhaustive, of the altarpiece exterior.
we will refer to other publications on the 18 The textile is impregnated with the
pictorial technique of the polyptych in chalk-glue mixture of the ground – both
the notes that follow. non-radio-opaque materials – which
9 Van Asperen de Boer 1979, p. 195, n. 71; explains why the imprint of the cloth is
Metzger, Steyaert 200 009, p. 167; Steyaert hard to see in the xrs.
20012, p. 121. 19 Verougstraete 20015, pp. 68–71.
10 See chapter 4. Examples of pre-Eyckian paintings
103
2. THE VAN EYCKS’ CREATIVE PROCESS
with textile strips glued to the panel: 30 If this is the case, the purpose of these
Melchior Broederlam, shutters of the lines, scored with a blunt tool – a kind
Crucifixion Altarpiece, Dijon, Musée of stylus or metalpoint – differs from
des Beaux-Arts, inv. 14200, see Currie that of the lines drawn in a dry medium
200009, p. 33; Coronation of the Virgin and that were detected in the panels of the
Annunciation, Antwerp, Royal Museum altarpiece exterior, where we think that
of Fine Arts, inv. 516, see Verougstraete metalpoint could have been used to lay
20015, pp. 325–26; Diest Last Judgement, down the initial drawing, before being
Brussels, Royal Museums of Fine Arts strengthened and completed in brush
of Belgium, inv. 4658, see Dubois et al. with a liquid medium. Postec, Steyaert
200009, p. 61; St Maurice Shrine, Namur, 200200, pp. 200
00–2006.
Musée provincial des Arts anciens 31 This point will be discussed in more
du Namurois, inv. 1500, see Deneffe, detail in the section ‘The dove, the halo
Peters, Fremout 200 009, p. 364; see also and the rays’.
‘Preparation of the picture support’ in
Von Baum et al. 20014, pp. 31–35; Serrat 32 Contrary to what is stated in Martens
20019, p. 71. 20019, p. 128.
20 Verougstraete 20015, pp. 412–16. 33 Van Asperen de Boer 1979, p. 194.
21 Skaug 200 006; Kühnen, Hern 20016. 34 Postec, Steyaert 200200, pp. 2007–21.
22 This hypothesis was suggested by Jean- 35 Postec, Steyaert 200200, p. 200
00 and n. 33.
Albert Glatigny (oral communication). 36 See the analyses in chapter 3, section
Two of the shutters – the Knights of Christ ‘Underdrawing’.
and the Hermits – show evidence of very 37 Jan van Eyck, St Barbara, 1437, Antwerp,
early repairs (in the sky of the Knights Royal Museum of Fine Arts, inv. 410.
of Christ and in the brown robe of the Van Asperen de Boer 1992; Billinge,
fi gure on the left of the Hermits), which Verougstraete, Van Schoute 2000 000; Postec,
we will discuss in detail in a moment. Sanyova 20016. The presence of washes has
It is diffi cult to determine whether these been noted in several works by Jan van
repairs and the application of the canvas Eyck; see Postec, Steyaert 200200, p. 245,
strip in the central panel occurred at the n. 53.
same time. The canvas does appear to 38 See chapter 3, section ‘Priming’.
have been placed on the central panel
before the ground was applied, while 39 See chapter 3, section ‘Priming’.
some paint layers of the wings had 40 Coremans 1953, p. 110, and pl. xxxvii-4;
already been executed at the time Van Asperen de Boer 1979, p. 2003,
of the repairs. pl. 69.
23 Van Asperen de Boer 1979; Périer- 41 Postec, Steyaert 200200, p. 2200; Billinge
D’Ieteren 20017. et al. 1997b; Von Baum et al. 20014,
24 As we will see in chapter 4, the execution pp. 47–57.
of these preparatory layers most likely 42 Van Asperen de Boer 1979, pp. 194 and
predates Jan’s intervention. 2001.
25 Van Asperen de Boer 1979, pp. 178–2001. 43 The nature of the copper pigment,
26 Postec, Steyaert 200200, p. 20000. verdigris, was confi rmed by analysis.
See chapter 3, section ‘Green’.
27 Postec, Steyaert 200200, pp. 20000–216.
44 Cross-section 10/82.
28 Coremans, 1953, p. 110 and pl. xxxvi;
Van Asperen de Boer 1979, p. 199, fi gs. 45 Kirby 20012, p. 272; Spring 20017, p. 17.
66a and b. 46 This oval shape, the lower part of which
29 Coremans counted 29 scored rays bulges out towards the bottom of the
(Coremans 1953, p. 110). It is diffi cult panel, conceivably follows the shape
to count the rays drawn with a liquid of the preparatory drawing, which
medium in the irrs, as they are not established the position of the spring
always distinguishable from the gilded (the black paint prevents the drawing
rays, because the mordant also absorbs from being visible in the irr).
infrared. Even so, 27 rays drawn with a 47 God Placing Adam in the Garden of Eden,
liquid medium can be counted in the area miniature from Guyart des Moulins,
directly around the dove. Bible historiale, Paris, Bibliothèque de
104
2. THE VAN EYCKS’ CREATIVE PROCESS
l’Arsenal, MS 50057, fol. 8; see Meiss 59 The green underlayer is also visible in the
1 974, pp. 339–42. barb along the right and bottom edges.
48 Limbourg Brothers, La Rencontre des 60 Van Asperen de Boer 1979, pp. 178–800.
Mages, in Les Très Riches Heures du duc de 61 Even though we could only make out this
Berry, Chantilly, Musée Condé, MS 65, underlying green layer by the naked eye
fol. 51. along the right edge in the Pilgrims.
49 Light pictorial layer consisting of lead 62 Unless the patch of grass dates from
white, ochre, verdigris, chalk and lead- the third stage, which would mean the
tin yellow. change occurred at a later date.
50 High-resolution ma-xrf maps could 63 This stratigraphy is confi rmed by cross-
not be made of the entire surface of the section 09/19, observations made by
painting. stereomicroscope and analysis of the
51 As we will see in due course, however, ma-xrf maps, xr and irr.
the present altar is the result of an 64 Philippot, Sneyers 1953, p. 83,
enlargement that might be non-Eyckian. mentioned ‘surpeints anciens dans la
52 The irr of the Fountain of Life in the végétation au-dessus des rochers’.
Museo Nacional del Prado (follower of 65 Only two layers of azurite and lead white
Van Eyck, 14400–500, inv. p1511) shows were detected in a few cross-sections,
several segments in the preparatory e.g. 12/003, but these were generally the
drawing of the octagonal fountain (Pérez result of samples taken from the edges of
Preciado 20018, p. 47, ill. 35). The arcs the panels, making the interpretation less
here appear to be clearly linked to the reliable.
basin. They might be construction lines
for the drawing of octagonal shapes, or 66 Campbell 20017, p. 258; Postec, Steyaert
perhaps the basin was initially round. 200200, pp. 33, 43, 49.
There is no trace of a similar dark 67 Given the relatively simple stratigraphy
underlayer or bulge in the irr, nor is of these passages, the sky is generally not
there any grass around the fountain. the preferred place from which to take
53 Van den Bremt, Van Crombrugge 200200, paint samples. We found two examples
pp. 252–53. in which the sky comprises two layers
containing azurite: Petrus Christus,
54 The scored lines used to position the The Lamentation (Brussels, Royal
stone basin became visible to the naked Museums of Fine Arts of Belgium, inv.
eye when the thick layers of varnish and 564), Kockaert 1995, p. 190 0; Jan Gossaert,
retouches were removed. There was no The Adoration of the Magi (London, The
generalized overpainting in this area, National Gallery, ng27900), Billinge et al.
merely a few localized retouches and 1997a, pp. 92–93. In a third example, the
oxidized varnishes. sky is built up in two layers, one based
55 Similar lines are also found in the on azurite and the other on ultramarine:
paintings of the altarpiece exterior, see follower of Rogier van der Weyden, Pieta
Postec, Steyaert 200200, pp. 2006 and 218. (London, The National Gallery, ng6265),
For other examples with compass marks Billinge et al. 1997a, p. 77.
in Netherlandish painting, see: Metzger, 68 We also note that a similar problem is
009, pp. 172–73 and n. 41;
Steyaert 200 located in the robe of one of the hermits
Steyaert 20012, pp. 132–33. in the right wing. See chapter 3, section
56 Due to abrasion of the light-brown ‘Zinc vitriol’.
layer, the grey underlayer is exposed in 69 The fi rst indigo-based layer is also
numerous places. missing. These investigations were
57 We know this was the new meadow, as performed using samples taken from
lines were scored in the fresh paint of and alongside the damage, in a zone
this fresh green pictorial layer in order to displaying other, more recent losses.
mark the position of the column. 70 Other indications, visible in the xr and
58 It cannot be ruled out, in our view, that the irr of the Joos Vijd panel, might
the brown glaze, the highlights and other indicate early damage: stains, indications
accents along the edges of the column of density differences along the joint at
and which overlap the surrounding the corresponding site, as well as in the
meadow are a non-Eyckian intervention. arch.
105
2. THE VAN EYCKS’ CREATIVE PROCESS
71 As we will see, this is not the only of the thick layer of sky that was
indication, see p. 69 and note 83 below. removed during the 20016–19 restoration
72 The ridge is split in places in the xr, treatment.
suggesting that this might be two 77 As stated in the ‘Note to the Reader’, we
different stages of execution, unless the use ‘Eyckian’ to refer to everything that
more weakly contrasting zones in the xr could have been executed by Hubert or
correspond with differences in thickness Jan, and ‘non Eyckian’ for everything
within the same layer, which is rich in not executed by the two brothers, which
lead white. It is hard to interpret the xr does not exclude the intervention of a
of the horizon, as the sky is composed of contemporary painter.
different layers of paint, each containing 78 Coremans, Loose, Thissen 1953, p. 112.
lead white and each of which might
deviate a little – but differently in each 79 For the sake of clarity, we have retained
case – from the horizon line or the the numbering of the building groups
outlines of the planned buildings. What used by Coremans, Loose, Thissen 1953,
is more, there are also several hills and p. 111.
mountains in this zone, some of them 80 Coremans, Loose, Thissen 1953, p. 112.
painted on top of each other, all of which 81 Van Asperen de Boer noted changes
also contain lead white. Regarding in the architecture that he situated
‘borders forming a ridge’, see also essentially between the preparatory
Van Asperen de Boer 1979, n. 76. drawing and the painting, although
73 This is what Coremans called ‘un double he detected the presence of dark layers
changement de composition’, Coremans, beneath the tower of Utrecht cathedral,
Loose, Thissen 1953, p. 112. which he interpreted more as a fi rst ‘dark
undermodelling’. Van Asperen de Boer
74 It should be noted that no varnish 1 979, pp. 187–88.
appears to be present between the fi nal
layer of azurite and the ultramarine layer 82 Coremans, Loose, Thissen 1953, p. 112.
(no intermediate varnish was detected 83 This is an additional indication that the
in any of the samples in which we fi rst layer of azurite in the sky belongs to
believe a third-stage layer to be present). the initial concept.
The application of the ultramarine 84 See chapter 1 (fi g. 1.21).
layer might have been preceded by
cleaning. There is no evidence, however, 85 The light-blue mountains in the
that Flemish painters systematically background of this city could belong to
varnished their paintings. A few rare the initial concept. They are similar to
traces of possibly original varnish or the small area of light-blue mountain
intermediate layers have been found visible to the left of the tower of Utrecht
on medieval paintings (Sauerberg et al. cathedral.
009, pp. 244–46; Dunkerton, Morrison,
200 86 In irr, some forms do not correspond
Roy 20017, pp. 275–78). However, with the current landscape, particularly
certain sources – albeit Italian – suggest to the left of the cypress. These forms
that distant blues were not necessarily probably correspond with the initially
varnished (Glanville 1995, p. 14; Postec planned landscape and indicate the
006, pp. 229–31).
200 possible presence of fi elds.
75 This ultramarine-based level is repeated 87 Coremans, Loose, Thissen 1953, p. 118;
in Coxcie’s 1557 copy, and might Van Asperen de Boer 1979, pp. 182–83.
correspond with Blondeel and Van 88 Van Asperen de Boer 1979, p. 197.
Scorel’s documented intervention 89 Identifi ed as angular Solomon’s seal,
(see Dubois 20017). For the dating of scented Solomon’s seal; see Van den Bremt,
the overpaint removed from the sky, Van Crombrugge 200200, pp. 210–11.
see also chapter 3, section ‘Soda-ash
glass’. 90 Van den Bremt, Van Crombrugge 200200,
pp. 1600–61, 218–19.
76 One of those losses is present at the
right side of the bell tower, in the sky 91 These are also visible in the copper,
and a leaf of the large fi g tree. The calcium and iron ma-xrf maps.
retouching of the leaf in this loss might 92 Possibly hazel; Van den Bremt,
be contemporary with the application Van Crombrugge 200200, pp. 146–47.
106
2. THE VAN EYCKS’ CREATIVE PROCESS
107
2. THE VAN EYCKS’ CREATIVE PROCESS
108
2. THE VAN EYCKS’ CREATIVE PROCESS
painting of it in the xr and ma-xrf iron the conservators would have found traces
and zinc maps. of them during the very many hours
135 Coremans, Loose, Thissen. 1953, p. 106, required to remove the sixteenth-
note 21 and pl. xxvi; Van Asperen de century overpaint over the entire sky.
Boer 1979, p. 1800. The ultramarine layer presented
enough small losses – or was suffi ciently
136 The underlying chaperon, which is translucent in places – to offer a view
slightly smaller than the current one
of the underlying layers.
and which leaves a reserve for the
crown, gives a high signal of copper, 147 See chapter 4.
corresponding with verdigris or azurite 148 Beenken 1933, p. 68; Beenken
blue, and of lead in the ma-xrf Cu-K 1 933–34, pp. 27–28; Panofsky 1 953,
and Pb-L maps. The painted crown, pp. 218, 225; Van Asperen de Boer
meanwhile, is readily legible in the ir 1 979, pp. 195–97, 200 00, 212; Dubois,
photograph, xr and ma-xrf iron map. Sanyova, Vanwijnsberghe 20017,
See chapter 3, fi g. 3.9. pp. 71–73.
137 This is the sixteenth-century overpaint 149 These samples were taken from the
that was further removed during the dove at the edge of its silhouette. See
current restoration campaign; see chapter 3, section ‘Wood ash with some
chapter 1. fern ash’. It is not impossible that a fi rst,
138 See fi g. 1.8. smaller version of the bird is present
139 Depoorter 200200, p. 52. beneath the current one. One of the
140 Van der Snickt et al. 200200, pp. 3–7. cross-sections was already described by
The authors have not been able to Coremans, Loose, Thissen 1953, p. 108,
determine, however, whether or not pl. xxxvii–2.
these changes are Eyckian. 150 Another detail worth noting is the
141 A dowel pin also caused damage to the presence of fragments of gilding below
Interior View in the altarpiece exterior; the upper white pictorial layer of the
Ketels, Glatigny, Augustyniak 200200, dove (Van Asperen de Boer 1979, p. 196).
p. 48; p. 500, fi g. 2.3 and n. 19. Damage Old gilding around the dove was
related to the presence of a dowel was removed during the restoration campaign
also noted in the Virgin Annunciate, just in the 19500s along with the clouds over
above the wing of the dove. the halo. Some fragments might remain
of this gilded background, which has now
142 Van Asperen de Boer 1979, p. 194. been removed, a few of these fragments
143 See note 29. are clearly trapped between two white
144 Van Asperen de Boer 1979, p. 199. paint layers of the dove.
145 Coremans (1953, p. 111) concluded that 151 See chapter 3 (section ‘Powdered glass’)
the system of ‘rayons gravés’ had been on the laboratory analysis regarding
abandoned. the addition of powdered glass to the
146 If any rays had existed beneath the pictorial layers.
ultramarine layer, it is highly likely that 152 See chapter 3, section ‘Paint additives’.
109
20
0 μm
3
The Challenges of a Complex Stratigraphy
from a Chemical Point of View
Introduction
The technical virtuosity and highly prized status of one of the world’s most
important large-scale artworks make the Ghent Altarpiece stand as a reference for
the study of the technique of early Netherlandish painting and of Hubert and Jan
van Eyck in particular.
The materials and techniques of Jan van Eyck and his contemporaries have been
thoroughly investigated and are relatively well understood. Since the fi rst and well-
known scientifi c study conducted in the early 19500s by Paul Coremans and his
team,1 our understanding of fi fteenth-century painting studio practice has been
considerably enlarged. In the following decades, the numerous technical studies
of Jan van Eyck’s paintings in Berlin, Bruges, London and Washington generated
supplementary insights that facilitated the interpretation of data obtained
during the current investigations of the Ghent Altarpiece.2 Also, recent research on
pigments and additives and their complex interactions with binders enables us to
contextualize our fi ndings. For instance, the intensive use of powdered glass and
zinc sulphate as additives in the Ghent Altarpiece panels, further discussed below,
ties in well with recent fi ndings in other early Netherlandish paintings.3
The large size of the polyptych presented a challenge to its material investigation.
The more than twenty-fi ve square metres of painted surface with a very turbulent
material history provide unprecedented opportunities for an in-depth study
during the time-consuming, meticulous work on its restoration. Indeed, studying
(facing page)
the painting technique requires linking visual observations, supported by X-
Adoration of the
Lamb: cross-section
radiography (xr), infrared refl ectography (irr) or ultraviolet (uv) photography, with
of the dove (10/18) the insights obtained from chemical analysis. The latter is performed with a wide
under vis, combined range of instruments and methods, which are subject to continuous innovation,
with back-scattered even during the course of the treatment. For instance, macro X-ray fl uorescence
electron sem image (ma-xrf) was used for the fi rst time for large-scale elemental mapping during
(see also 3.18f–g). the fi rst restoration phase,4 while, in the second phase, the main novelty was the
111
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Layer build-up
The cross-sections, prepared from the sample set taken during the 19500–51 treatment
of the altarpiece before removal of the overpaint, often revealed a very complex
stratigraphy with a number of layers both larger and more variable than usually
encountered in cross-sections of early Netherlandish paintings. A certain variability
was also observed in other paintings by Jan van Eyck (e.g. the Annunciation in the
National Gallery of Art, Washington, and the Arnolfini Portrait in the National
Gallery, London6) but seemed to correspond to the modelling, shading and/or
changes in the composition. However, in the Ghent Altarpiece the unusual number
of layers (sometimes up to ten under the fi rst identifi able varnish) could also have
112
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Preparatory layers
Ground
113
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Group Characteristics of the priming oil-based layers Panels in which these layers are encountered
Composed of one or two translucent layers (overall Interior: lower register (Knights of Christ,
thickness 10–15 µm) containing particles of calcium Adoration of the Lamb, Hermits and Pilgrims)
A carbonate, lead white, carbon black. Sometimes a few Interior: upper register (Singing Angels, Virgin
red particles such as red earth, vermilion or minium are Enthroned, Deity Enthroned, St John the Baptist
present (mainly under the green and blue areas). Enthroned and the Angel Musicians)*
Quite thick (10–300 µm), greyish, semi-translucent layer Exterior: upper register (Archangel Annunciate,
containing particles of bone white, calcite, lead white City View, Interior with Lavabo and Virgin
B
and variable proportions of coal black. Sometimes a few Annunciate)
particles of vermilion or red earth are encountered. Interior: upper register (Adam)*
Very thin (1–3 µm), unpigmented transparent layer, Exterior: lower register (panels Joos Vijd, John the
C in a few cases not visible (absorbed?) or exceptionally Baptist, John the Evangelist and Elisabeth Borluut)
located below the underdrawing.
Underdrawing
114
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
(fi g. 3.16b). This exceptional thickness facilitated the atr-ftir (Attenuated total
refl ection – Fourier transform infrared spectroscopy) identifi cation of proteins used
as a binder, thus confi rming the underdrawing’s aqueous character.
Priming
] ]
] ]
50
0 μm 50
0 μm 50
0 μm 50
0 μm
115
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
carbonate19 and carbon black. In several cross-sections, mainly those extracted from
green or blue areas, a few red particles identifi ed as vermilion, red lead or a red earth
were found as well. A priming layer with similar characteristics was detected in
the panels of the upper interior register, the Singing Angels, the Virgin, Deity and St
John the Baptist Enthroned, and the Angel Musicians in a preliminary examination of
the old samples. Interestingly, the preliminary analysis of a sample from the Adam
panel suggests that it rather belongs to group B (see below) than to group A.20
This group A in priming could relate to the fl esh-like, beige primuursel, discussed
by Karel van Mander in his Schilder-boeck (16004). The author concludes that this
was the standard mixture of materials (or standard painting technique if he does
not mention the pigments) of his fi fteenth- and sixteenth-century predecessors,
explicitly including Jan van Eyck.21
Group B (exterior: upper register) includes the Archangel, the City View, the
Interior View and the Virgin Annunciate (fi g. 3.3). The layer is signifi cantly thicker,
up to 300 µm, more pigmented and therefore less translucent, but the drawing
is still visible. In certain areas the priming layer serves as an underpainting. In
particular, the blue and pinkish glazes of the wing of the Archangel or the rose glaze
in the prophet Zechariah’s drapery (fi gs. 3.3b and 3.16) were applied directly on
top of the greyish priming layer.22 Here, the layer holds particles of calcined bone,
hydrocerussite, cerussite, calcite, variable proportions of coal black and, only in the
Virgin Annunciate panel, some vermilion and red earth.
The fi nding of calcined bone in the priming of fi ve Ghent Altarpiece panels was
surprising, as technical studies rarely report this material in West European panel
paintings. To the best of our knowledge, only two other cases of similar priming
were reported in early Northern European painting. One is Memling’s large-size
ensemble of God the Father with Singing and Music-Making Angels (Royal Museum
of Fine Arts Antwerp, inv. 778–7800)23 and the second is the Norwegian frontal
Tresfjord produced in the second quarter of the fourteenth century.24 Calcined
bone was also found in the gilded pastiglia letters on the frame of the Crucifixion
(Metropolitan Museum of Art, New York, acc. 33.92ab) attributed to Jan van Eyck
and a workshop assistant,25 as well as in the gilding mordant and in the tin foil
adhesive of the pre-Eyckian Scenes from the Life of the Virgin (Royal Museums of Fine
Arts of Belgium, Brussels, inv. 4883) dated around 1400 00.26
]
]
] ]
50
0 μm 50
0 μm 50
0 μm 50
0 μm
116
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
40 μm 40 μm
3.4b 3.4c
Cerussite Hydrocerussite
3.4d 3.4e
Hydroxylapatite Pb-carboxylates
3.4a 3.4f 3.4g
Fig. 3.4a–g. Archangel: detail of the prophet Zechariah’s red sleeve. Location of sampling (a), paint cross-section (13/22)
in vis (b) and in uv light (c). Spatial distributions of cerussite (d), hydrocerussite (e), hydroxylapatite (f) and lead
carboxylates (g) within the cross-section visualized using sr-µ-xrpd. Arrows indicate the priming layer in b–g.
In medieval recipe books for artists, burnt bone is usually recommended for the
preparation of a suitable surface for metalpoint drawing.27 However, it is also proposed
as an additive for improving the drying of oil. Two fi fteenth-century German
manuscripts, the Strasbourg and the Tegernsee manuscripts,28 include bone white
in recipes for preparing an oil-based paint medium, a gold varnish and a mordant
for gilding. But the most relevant recipe for priming is found in the fourteenth-
century Icelandic manuscript known as Likneskjusmiå: ‘… you should grind burnt
bone with oil and this is called underwhite. It should be spread on the whiting with
a paintbrush on those places where you intend to apply other colours …’.29 This
description fi ts quite well with what is observed in the upper register of the exterior
Ghent panels where the priming is simultaneously fulfi lling the role of underpaint.
The chemical composition of this particular kind of priming was studied in the
sample taken from the sleeve of the prophet Zechariah’s drapery using synchrotron
radiation-based (sr) analysis.30 The sr-µ-xrpd mapping of the cross-section from
this area (fi g. 3.4) demonstrated the presence of a relatively high concentration of
hydroxyapatite, from calcined bone, evenly dispersed in the priming.
A small fraction of calcite was also found, most likely part of the calcined bone or
stemming from the addition of chalk or carbonated quicklime. The latter is indeed
recommended in a fourteenth-century manuscript for the preparation of a sun-
thickened leaded oil.31 The addition of highly alkaline quicklime to oil containing
lead white would catalyse the saponifi cation of triglycerides and the formation of
metal carboxylates which accelerate oil drying. Lime, calcined bone and also zinc
vitriol, would act as secondary (auxiliary) driers, reinforcing the drying property of
oils containing lead- or manganese-based primary driers.32
117
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
50
0 μm 50
0 μm 50
0 μm 50
0 μm
Finally, sem-edx mapping has shown that lead is homogeneously dispersed in the
binder and sometimes associated with discrete particles of lead white. sr-µ-xrpd
(fi g. 3.4) also revealed the presence of two lead carbonates, basic hydrocerussite
(2PbCO3.Pb(OH)2) and neutral cerussite (PbCO3). At the moment, it is not clear
to what extent these lead carbonates were added to the starting mixture, or were
formed by remineralization of lead carboxylates; both hypotheses are conceivable.
As demonstrated by Cotte, lead carboxylates could have been formed during
boiling of the oil or ‘maturation’ by sun-thickening and it is well known that lead
carboxylates are also spontaneously formed during the ageing of a paint fi lm by the
interaction of the binder and lead white.33
In group C are placed all the panels of the lower register of the exterior shutters:
Joos Vijd, John the Baptist, John the Evangelist and Elisabeth Borluut. The isolation layer in
these paintings is very thin (ranging from 1 to 3 µm), rather unpigmented (containing
only a small amount of chalk), and thus completely transparent. Thanks to its strong
fl uorescence, it can readily be recognized when the cross-sections are observed in a
microscope under UV illumination (fi g. 3.5c). As it is almost unpigmented, it might
have penetrated into the underlying layers. It is thus hardly detectable in such cases.34
Paint layers
Palette overview
The Eyckian palette used for the Ghent Altarpiece was not extensive; it included
around twenty pigments commonly found in fi fteenth- and sixteenth-century
Northern European painting.35 In general, most of these pigments were correctly
identifi ed in the previous studies of samples taken from the polyptych in 19500–51.36
The list of materials previously found in the panels’ original paint layers was Fig. 3.6. Pigments
complemented with pigments and additives. Thus coal black, bistre, kermes lake, identifi ed in the
orpiment, bone white, white vitriol or colourless powdered glass were added to the Ghent Altarpiece
existing list which included lead white, chalk, carbon black, azurite, ultramarine, panels studied
indigo, verdigris, malachite, lead-tin yellow, yellow ochres, vermilion, red earth between 20012 and
and red lead (minium) (fi g. 3.6). 20019.
118
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
White Lead white Only opaque white pigment used, rarely used alone, more often
for modelling
Blue Ultramarine Less frequent than azurite in blue and violet areas
Lead-tin yellow type 1 Frequently used for bright yellow tones and in green mixtures
Red earth (hematite-based earth pigments) Not used alone, found in small amounts in red glazes, in
painted brocade motifs, green verdigris-containing layers, grey
Red underlayers of sky, sometimes in priming
Kermes lakes (containing traces of lac-dye) Standard red pigment, alone or in mixture with blue or black
Red lead Not found in paint layers, only in priming and mordants
Coal (sulphur-rich, containing some Main black pigment, used in underdrawing and shadowing
chlorine)
Black and Charcoal Used to a much lesser extent than coal
brown
Bistre from soot Detected in a few brown paint layers
Powdered potash-lime glass Used in red and green glazes and in the black lettering of
frames
White vitriol (zinc sulphate) Frequently used in small amounts, mainly in red, brown and
green layers
Additives
(less than Quartz Used in red, pink and green glazes, in priming and in blue,
3%) rose and brown opaque layers, also present as ultramarine
and impurities
extenders
Bone white (hydroxylapatite-based) Extender in type b priming layers, likely as an additive in
brown and green layers
Chalk (calcium carbonate) Extender of ground and priming, as an additive in red, green,
brown glazes, grey
3.6
119
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Pb-M Pb-L
120
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
White
Lead white (basic lead carbonate) is one of the most widely used pigments in the
polyptych. It was added in all opaque layers and is present in quite low proportion
in both the (semi-)transparent and the priming layers.38 It was also used in later
interventions. Historically, lead white was produced artifi cially by the corrosion of
metallic lead in vinegar, leading to the formation of lead acetate, which is transformed
into lead carbonate by carbon dioxide. The resulting material is mainly composed
of two crystalline phases of lead carbonate, hydrocerussite (HC, 2PbCO3·Pb(OH)2)
and cerussite (C, PbCO3). The synthesis and purifi cation methods applied during
pigment processing impact not only the pigment’s physico-chemical properties and
quality grade but also the hc/(hc+c) ratio value. Given that it can be measured
with X-ray diffraction techniques, this relation can thus be used to categorize
different types of white lead found in paintings.39
Considering the above, the ma-xrpd40 mapping of the Ghent panels in the
areas containing lead white allowed some interesting observations to be made.
For example, in the background of the Hermits panel, the lead white used for the
female fi gures contains a higher proportion of cerussite than the one found in the
group of male fi gures and mountain rocks in the background, which are richer in
hydrocerussite (fi g. 3.7). This result suggests that (at least) two different types of
lead white were used. It is not unthinkable, therefore, that the female fi gures and
the groups of hermits were not painted at the same time and thus not necessarily by
the same artist, unless the use of two different types of lead white was a deliberate
choice of the painter of this scene. However, this problem is more complex than
just the determination of the quality grade of lead white using the hc/(hc+c) ratio.
Frequent detection of plumbonacrite, a third lead carbonate that can be ascribed to
the carbonation of lead oxide (massicot or litharge) and lead carboxylates formed by
the interaction of an oily binder and lead pigment, complicates the interpretation
of the hc/(hc+c) ratio.41
ma-xrf scanning allows us to extract two distribution images for the element
lead (Pb) from the data cube, the Pb-L and Pb-M emission lines, each providing
different information. The map of the high-energy Pb-L emission lines reveals the
distribution of lead from underlying layers, showing their cracks and losses, while
the Pb-M map corresponds to the lead from the paint surface. The combination
of these two chemical maps allows us to examine the Eyckian painting technique
with some depth-selectivity. As shown in the illustration (fi g. 3.7), all faces and the
rocky mountain were underpainted with lead white and the modelling with fi nal
colours was realized in the upper layer.
Blue
Azurite The use of azurite as the main pigment in the blue areas in the altarpiece
panels was largely confi rmed by the ma-xrf copper maps. This pigment was
employed either alone or in admixture with red lakes to create shades of violet.
Inspection of cross-sections demonstrates its occurrence, both in the fi nal paint
121
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
3
2
1
50
0 μm
3.8b
3
2
1
50 μm
3.8a 3.8c
Fig. 3.8a–c. Adoration of the Lamb: detail of a velvet cloth of gold of a confessor. The white arrow indicates the sampling
location (a); paint cross-sections (10/76) under vis (b) and uv (c) light showing the layer structure: azurite-containing
underlayer (1), followed by ultramarine-containing brownish (2) and purple (3) glaze layers. The golden motifs were
painted with lead-tin yellow type I (4).
layers and in the underlayers (fi gs. 3.8 and 3.10). The pigment is commonly
intermixed with lead white, as observed in the skies, where it was applied in two or
three layers over an underlayer containing indigo and combined with incremental
proportions of lead white towards the horizon.
In the draperies, it was more often found underlying the ultramarine blue or
violet glazes. For instance, quite coarser azurite (ca. 200 µm particle size) was found
as an underlayer of purple and brownish glazes in the illusionistic representation of
the changeant or shot42 velvet cloth of gold fabric of one confessor in the Adoration
of the Lamb (fi g. 3.8).
122
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Sibyl’s and Zechariah’s headwear. The same goes for the interior shutters, where
ultramarine was only used in several small areas. In the Knights of Christ panel, the
jewels of the crowns and horse harnessing and the left knight’s helmet painted with
an ultramarine blue are considered to be original. However, the ultramarine areas
on this panel, such as the purplish velvet drapery covering the painted brocade
of the knight on the extreme left (fi g. 2.66), the large headgear of a knight in
the background covering a smaller, azurite-containing headgear with a crown (fi g.
3.9), and the mountains painted on top of the sky are probably later interventions.
A small amount of ultramarine was also detected in the mixture with red lake in
the dark purplish red of St Anthony’s drapery (fi g. 3.15).
A great deal of ultramarine was found in the Adoration of the Lamb, where
understanding the layer structure was most problematic. Consequently, the potential
differentiation of various ultramarine types was thought to be useful as a supporting
argument for the recognition of Eyckian and non-Eyckian interventions.43 Two
criteria for differentiating between them were explored: the distribution of the
different sizes of ultramarine particles and their elemental and mineralogical
impurities. Although it is too early to draw fi nal conclusions, a few tendencies
have been observed. For instance, the Eyckian ultramarine is more fi nely ground
Lazurite Nepheline
123
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
6 6
5 5
4 4
3
3
2
2
1
1
50 μm 50 μm
124
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
The nature of the binding medium for the Eyckian ultramarine blue is another
important issue related to this pigment. Based on various microchemical staining
tests performed on a sample taken from the Virgin Enthroned drapery, Coremans
and Kockaert concluded that ultramarine in the Ghent Altarpiece was applied using
an aqueous medium.45 This conclusion had already been questioned several years
ago, when progress in understanding of pigment–binder interactions is led to the
reinterpretation of the old analysis results.46 The atr-ftir analysis performed on
the ultramarine-containing layer of a sample from the Erythraean Sibyl’s turban
(fi g. 3.11) shows the absence of characteristic signals attributable to proteins (e.g.
amide I and II bands). Instead, it displayed the thorough presence of weddellite,
a calcium oxalate (CaC2O4·2H2O). These results are consistent with the published
observation on the alteration of oil-based binders in other works by Van Eyck.47
Extracted sp.
Arbitrary Units (A.U.)
Ca-oxalate
Gelatine
20 μm 1800
00 1700
00 1600
00 1500
00 1400
00 1300
00 1200
00 1100
00 1000
000 9 00
Wavenumber (cm-1)
3.11a 3.11b 3.11c 3.11d
Fig. 3.11 a–d. City View: detail of the Erythraean Sibyl’s turban sampling location (white arrow) (a). The paint cross-section
(14/003) under vis light (b) shows a detail of the area analysed by atr-ftir. Results obtained show the distribution of metal
oxalates (c) in the topmost, ultramarine-containing layer. The ftir spectrum (black) extracted from this layer (dotted black
square, c) shows characteristic calcium oxalate vibrations. Calcium oxalate (blue) and gelatine (red) reference spectra are
included for comparison (d).
125
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Green
The green pigment identifi ed in all green areas is the organo-copper pigment
verdigris. The identifi cation of verdigris is not straightforward, as its chemical
composition is not well defi ned.52 Produced by the corrosion of metallic copper or
its alloys exposed to the vapour of various organic acids (mainly acetic acid), the
composition of this pigment is indeed variable. It can include different copper salts
(acetates, chlorides, citrates, carbonates, formates, oxalates), of which copper acetate
is considered the main reference molecule.53 Its composition is not infl uenced only
by the raw materials used, but also by the purifi cation process (often in vinegar)
and the formulation of the paint mixture (binder, additives, etc.). Because of its
chemical composition, it reacts with the carboxylic acid moieties in oils and resins
resulting in green carboxylates and resinates, which catalyses the autoxidation of
oil. Because of its resulting drying properties and its aesthetic qualities, verdigris
was the green pigment preferred by early Netherlandish painters.
Cu-K Zn-K
Zn - oxalates
20
0 μm Cu-oxalates
1800
00 1700
00 1600
00 1500
00 1400
00 1300
00 1200
00 1100
00 1000
000 9 00
3.12d 3.12e
Wavenumber (cm-1)
3.12f
Fig. 3.12a–f. Pilgrims: detail of the cypresses and palm tree behind horizon (a) corresponding to the area imaged
by means of ma-xrf. The white arrow indicates sampling location (a). ma-xrf images showing the elemental
distribution of copper (Cu-K) (b) and zinc (Zn-K) (c). The paint cross-section (12/13) (d) from the brownish
cypress contains two layers: the light blue lower one corresponds to the sky layer and the brownish upper one
is related to the cypress. The atr-ftir spatial distribution map shows the presence of metal carboxylates (e),
in both layers. Moreover, copper and zinc oxalates were also detected in the brownish upper layer (dotted black
square (e) as shown in the extracted ftir spectrum (black). Zinc (blue) and copper (red) oxalates reference spectra
are included for comparison (f).
126
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
In the Ghent Altarpiece, verdigris is present in all green areas, including the
vegetation and draperies, and was used in all superimposed layers. It was often mixed
with minor amounts of other pigments in the modelled underlayer and used pure
in the green glazes. Sometimes, a thin yellow glaze was recognizable over the green
one, probably to enhance and saturate the colour with a yellow dye, possibly saffron
or barberry, as recommended by the Strasbourg Manuscript.54
Verdigris is nearly always found in admixture with small amounts of other
pigments, additives and/or dyes that are often diffi cult to identify. It is clear in many
cases that these were added intentionally, as was the case with the aforementioned
ultramarine or indigo blue, but also with lead white, lead-tin yellow, yellow and
brown ochres, orpiment, carbon black, yellow lakes or glass and quartz. Copper
chlorides, however, which were also quite systematically found in the Ghent
Altarpiece’s green samples, were not a deliberate addition. These compounds were
more probably formed as a by-product of the production of verdigris prepared
following one of the recipes using sodium chloride (salt). That ingredient was
used to slow down the corrosion of copper by the organic acid (e.g. acetic acid),
allowing the latter to penetrate deeper into the copper plate. Besides copper
chlorides, copper oxalates and copper formates were also detected in the verdigris
replicated from historical recipes.55 Their presence in the historical samples should
therefore not be automatically considered to result from the interaction of binder
and verdigris, as is often proposed also for the copper carboxylates. Nevertheless,
this phenomenon is still not well understood and needs to be further investigated.56
In the Ghent Altarpiece panels, copper carboxylates, copper formates and copper
oxalates (moolooite) were identifi ed in several green zones by atr-ftir and/or
ma-xrpd. Through ma-xrpd results it was observed that moolooite is present
in higher concentration in the green-brownish vegetation areas where the lead-
tin yellow or a zinc-containing additive were codetected.57 This was, for instance,
the case in the green areas with vegetation along the horizon (cypresses, shrubs
and trees) in the Adoration of the Lamb, the Knights of Christ, the Hermits and the
Pilgrims (fi g. 3.12a). The phenomenon can be illustrated by the ma-xrf results for
copper and zinc of one such area on the left side of the Pilgrims panel (fi g. 3.12b–c),
correlated with the atr-ftir mapping results of a cross-section sample taken in
the cypress. Results obtained with the latter technique revealed metal carboxylates
in both layers constituting the pictorial system (fi g. 3.12e): lead carboxylates in
the lower, light blue sky layer; and copper and zinc carboxylates and oxalates in
the zinc-containing brownish transparent upper layer, related to the cypress tree
painted over the horizon (fi g. 3.12d–e).
It is worth mentioning that no copper oxalates (moolooite, Cu(C2O4)·nH2O, n<1)
were identifi ed by ma-xrpd in the azurite-containing areas ((Cu3(CO3)2(OH)2)).
However, these metal oxalates were found in the green-yellowish malachite-
containing ((Cu2(CO3)(OH)2)) drapery of a rider in the Knights of Christ (fi g. 3.13),
but in signifi cantly smaller amounts than in the verdigris-containing greens.
Interestingly, this was the only green region, probably from the Eyckian stage,
where the pigment malachite was positively identifi ed during the present
campaign.58 Unfortunately, as mentioned above, no samples concerning this area
were available. Thus, the position of malachite in the paint layer structure and the
127
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
In the green areas, verdigris was often in admixture with organic brown-yellow or
brown-black pigments. These pigments, together with red and yellow lakes, hold
pride of place in the Van Eyck brothers’ palette used for the Ghent Altarpiece. As they
form a translucent layer when mixed with oil, their use cannot be dissociated from
the Eyckian painting technique, which is characterized by the superimposition of
translucent layers. However, the organic pigments are also the most problematic
to identify and their detection and localization are often merely assumed on the
basis of visual observation of a colour and/or the identifi cation of additives or
extenders. These pigments share the common feature that they are poor driers and
thus do not offer an appropriate paint consistency. Accordingly, they need to be
associated with a drying agent and/or an extender. Their organic constituents can
only be identifi ed using chromatographic methods, such as hplc for the dyestuffs
and gc-ms for the specifi c markers in the soot-brown (e.g. guaiacol and syringol
derivates).
Carbon-based brown and black A soot-dark brown pigment, called bistre, was
identifi ed with py-gcms in two dark brown paints inside the frame of the Just
Judges. The recovered paint, supposed to be an Eyckian material, fl owed and
solidifi ed into a gap between the panel and the frame. One sample originated from
a dark brown area, presumably featuring rocks, the other from the green-brownish
128
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
50
0 μm
3.14a 3.14b
Fig. 3.14a–b. Hermits:
detail of the far-
left hermit’s brown
garment – sampling vegetation on the left side of the painting.59 Bistre is a dark brownish carbon-based
location (a). Back- pigment extracted from chimney wood soot, referred to as caligo and fuligo in the
scattered electron manuscript of Jehan le Begue or gelutert rus in the Strasbourg Manuscript.60 Despite
sem image of the its hydrophobic, slightly resinous character, bistre is more often recommended
paint cross-section for watercolour and for washes and ink drawings. However, it has been identifi ed
(11/k) (b) shows large in medieval paintings from Cologne, in a brown oil-based paint containing also
coal-black pigment verdigris and ochre, similar to what has been found in the Ghent Altarpiece panels.61
particles.
Bistre was cheap and easy to make but remains diffi cult to identify. It is therefore
likely that it was more frequently used in fi fteenth-century paintings than reported
in technical studies.62
Coal black is another carbon-based pigment frequently detected in the Ghent
Altarpiece samples. It is recognizable in cross-sections thanks to the sharp edges
of its particles typical for a ground mineral rock, and shows a high concentration
of sulphur with chlorine, calcium, aluminium and silicon-containing impurities.
As already mentioned, it was used in the liquid underdrawing (fi g. 3.16) and was
frequently detected in the priming layers, where it was fi nely ground (fi gs. 3.2,
3.3 and 3.5). In the paint layer, coal was systematically observed in dark brown
areas, such as the drapery of the far-left Hermit (fi g. 3.14b), as well as in the black
underpaint of the dark green vegetation areas in the central panel and the interior
shutters. Likewise, coal black was added in some dark green and dark red glazes. In
short, it is the most used black pigment in the panels studied so far.
Charcoal, which could sometimes be mistaken for coal, is only sporadically
present in the Ghent Altarpiece samples. No bone black has been detected.
Red lakes Among the red pigments, lakes are dominant in the Ghent Altarpiece panels,
as they are present in all nuances of red, pink, violet and some dark brown colours.
The range of colours that can be obtained by the superposition of red lake-containing
glazes over opaque underlayers or even, in the case of the outer Annunciation panels,
over the tinted priming layer, is quite large indeed. For instance, at least seven
colour nuances are created with red lake pigments in the draperies and headdresses
of the fi gures in the foreground of the Adoration of the Lamb.
129
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
130
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
7
6
5
4
3
Priming 50
0 μm
Proteins Me-carboxylates
3.15a 3.15b 3.15c 3.15d
Fig. 3.15a–d. Hermits: detail of St Anthony’s purple garment sampling location (indicated by white arrow) (a).
The paint cross-section (11/12) under uv illumination (b) shows the squares matrix representing the atr-ftir analysed area.
The enabled analysis visualization of the spatial distribution of proteins (c) and metal carboxylates (d) in the cross-sections.
lighter areas (2, sometimes 3 layers). In some cases this number is even higher, as
can be seen in St Anthony’s dark purple mantle in the Hermits panel (fi g. 3.15b).
The paint structure in this area shows seven red glazes containing the expensive
kermes lake. This extensive and seemingly redundant pictorial construction might
be an indication that the work was executed in two stages.
Interestingly, some differences can be observed between the upper two and lower
fi ve layers. For instance, the dark purple colour was obtained by the addition of
coarser coal black in the lower layers whereas, in the uppermost layer, ultramarine
was employed instead. Likewise, the lake from wool shearings was detected mainly
in the lower layers as indicated by a strong protein signal observed in the atr-ftir
mapping (fi g. 3.15c). In the upper layers, the lake(s) seem(s) to be prepared from
dyed silk. It seems likely that lower paint layers were applied in a single stage of
the creative process while the two upper layers were added in a later stage.
On the exterior of the shutters, lakes were used to a large extent for Joos Vijd’s
red and Elisabeth Borluut’s pink drapery in the lower register, and in the Archangel’s
wings, in the Erythraean Sibyl’s drapery and in the prophets Zechariah’s and
Micah’s clothes, in the upper register. It is worth noting that the layer structure in
the red and pink areas of the upper register is much simpler than in those of the
lower register, clearly denoting different painting techniques. The red and pink
glazes in the upper register were applied in one or two layers, directly on the grey
priming (fi g. 3.16b), whereas those in the donor portraits, which are more vivid,
present a much more complex layer structure with three or four glazes applied
on an opaque underlayer. The underlayer of the bright red robe of Joos Vijd is
cinnabar-based, while that of the pink drapery of Elisabeth Borluut is red lake-based.
Another difference between the painting methods in the upper and lower exterior
panels consists in the use of additives: in the donor portraits the glazes contain
powdered glass, while zinc sulphate was detected with sr-xanes in the pink
131
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
S1
S2
S3
S4
ZnSO4.7H2O
CaSO4.2H2O
L-cysteine
L-cysteine
2.470
0 2.480
0 2.49 0 2.500
00 2.510
0 2.470
0 2.480
0 2.49 0 2.500
00 2.510
0
Energy (keV) Energy (keV)
3.16e 3.16f 3.16g 3.16h 3.16c 3.16d
Fig. 3.16a–d. Archangel: detail of the prophet Zechariah’s pink drapery. The white arrow indicates the sampling location on
the irr detail showing the underdrawing lines (a), paint cross-section (13/200) in uv illumination (b). sr-xanes S-K edge
spectra originating from the top pink layers (points s1 and s2) (c), and from the underdrawing (s3, s4) and the ground
(s5) (d), compared with the references for the thiol functional group (L-cysteine) and sulphate function (ZnSO4.7H2O and
CaSO4.2H2O) (c, d). Distributions of sulphur (S-K, e), chlorine (Cl-K, f), lead (Pb-M, g) and phosphorus (P-K, h) within
the cross-section visualized using sr-µ-xrf.
glazes applied to depict the prophet Zechariah’s clothes (fi g. 3.16c–d).69 It should
be stressed that these two additives, which were expected to accelerate the drying of
oil70 (together with others detected in the Ghent Altarpiece samples, such as calcined
bone, chalk, copper vitriol, quartz and small amounts of ‘good’ drying pigments,
such as verdigris or lead white), play an important role in the understanding of
the Eyckian paint formulation and behaviour during the grinding, application,
drying and ageing processes. For this reason they were investigated with particular
attention, as explained below.
Paint additives
The brushstrokes of Jan van Eyck are characterized as quick and precise,71 and it
is clear that his paint mixture needed to have an appropriate consistency, neither
too viscous nor too liquid. Besides the binder and pigments, painters had at
their disposal a number of materials to modify the paint’s rheological and drying
properties. They were usually added in small amounts to modify the oil binder,
such as zinc or copper vitriol, calcined bone, lime or litharge. However, they could
also be added in quite a large proportion like chalk, quartz, starch or powdered
glass. Various modifi ers were encountered and systematically identifi ed in the Ghent
Altarpiece paint samples, but only two of them, glass particles and zinc vitriol, have
so far been studied in greater detail.
132
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Powdered glass
Powdered glass has been found at the Eyckian level in the red and pink glazes of
the two donor portraits and in the black paint of the letters of the quatrain on
the exterior grisaille panels. In the interior panels studied so far, very small glass
particles have been found in several of the Eyckian vegetation greens. Likewise,
larger glass particles of slightly different composition were identifi ed in the
Adoration of the Lamb, in the white paint layers of the dove and in the mordants for
some of the gilded rays surrounding it.
It should be noted that, ever since the 19500s, questions have been raised about
the dove because of its poor pictorial quality, uncommon for Van Eyck’s work.
Consequently, it has been hypothesized that it was added later by another painter,
when the altarpiece was fi nished, as discussed in chapters 2 and 4. Colourless glass
powder has also been found in later fi fteenth- and sixteenth-century interventions.
At these levels, the glass additive was used even more extensively than in the
Eyckian stage. These fi ndings correspond well with the evolution of painting
practice in the late fi fteenth and early sixteenth centuries. Indeed, glass powders
are recommended in the technical documentation of this period as siccative agents
in preparation of boiled oil and as an additive for grinding poorly drying pigments
such as red lake, indigo and lamp black in oil.72 However, this function is debatable
and has been widely discussed in recent scientifi c communications.73
In the medieval and renaissance periods, glass was manufactured by melting
sand or quartz pebbles with one or more alkali-rich fl uxes, which had important
effects on its chemical composition. In Northern Europe, the fl ux could have been
composed of various types of potassium-rich wood or plant ashes, while sodium-rich
coastal halophytic plants were employed in the Mediterranean area.74 A systematic
study of the elemental composition of the glass particles found in London’s
National Gallery paintings by Marika Spring has shown that calco-potassic glass
made with potassium-rich wood or fern ash was more common in Netherlandish
paintings. Within this group of paintings, various compositional categories of
glass were identifi ed, mainly based on the proportions of calcium and potassium
oxides (CaO/K2O): (a) wood ash (CaO/K2O between 1 and 2); (b) wood ash with
some fern ash (CaO/K2O below 1); (c) wood ash–lime (CaO/K2O above 3.5); and
(d) wood ash-lime with high lime/low alkali content (CaO/K2O between 5 and 8).75
Similar glass families were found in the samples taken from the Ghent Altarpiece.
The correlation between the category of glass and the stratigraphic position of the
layer in which the glass particles were found permitted us to propose a hypothesis
on the chronology of their use. The results are summarized in the table (fi g. 3.17),
followed by a short discussion of each category.
Wood ash
The particles made of wood ash-type glass were found in many samples at the
Eyckian level, although not systematically. The use of powdered glass at the
original level was indeed quite moderate in comparison with the overpaints. It
133
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Non-Eyckian addition (?) Wood ash with some fern ash (?) Dove and mordant for one of the gilded rays
Late 15th-century–early
1 6th-centuryoverpaints, Wood ash Red glaze in the Archangel’s wing
between 14700 (?) and 1557
1 6th-century overpaints, Red and pink glazes and fi llings of losses in major
Soda ash
before 1557 16th-centuryoverpaints
3.17
Fig. 3.17. Types
of glass particles
was found in the red, green and black layers in only four out of the twelve restored found in the twelve
panels and in four frames. restored panels of
A low sodium and high potassium and calcium oxides composition was observed in the Ghent Altarpiece
all samples. The proportion of the two latter oxides ranges from 1 to 2 (CaO/K2O and hypothetical
~1–2). The glass is equally rich in magnesium and phosphorus oxides. chronology of
In most cases, the glass was fi nely ground, with particle sizes varying from 3 their use.
to 15 µm, and down to even 2–5 µm in green glazes. Wood ash glass particles
were used mainly in the red and pink glazes of the two donor portraits, Joos Vijd
and Elisabeth Borluut, in the closed polyptych. A larger amount was found in the
Joos Vijd panel, while a smaller proportion was encountered in the pink drapery
of Elisabeth Borluut. It is interesting that the wood ash glass particles were rarely
found in the red clothes of the fi gures depicted in the lower register of the open
polyptych. They were found only in the red cloak of the bearded and bald prophet
in the lower left group of the Adoration of the Lamb. In this panel, the wood ash
glass particles were also detected in the verdigris-containing greens, and they were
similarly found in the greens in two other restored panels of the open polyptych,
the Knights and the Pilgrims. The black layers in which the powdered colourless
glass was also found were all sampled from the lettering on the frames of the
panels of the Knights, the Virgin Annunciate and the Archangel, and the St John the
Baptist.
Interestingly, it was detected in one sample at the level of an overpaint on the
Archangel Gabriel’s wing, where the red glaze with wood ash glass particles was
applied over an original red glaze covered by two varnishes. The top-most varnish
134
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
contained the particles of wood ash lime (high lime, low alkali [HLLA]) glass.
This type of glass frequently occurs in the paintings dated after 14700. Unlike other
Eyckian red glazes from the upper register of the exterior, the original red glaze
under this varnish does not contain any powdered glass.76
The type of glass containing wood ash with some fern ash was identifi ed in the
Adoration of the Lamb, in the thick upper white layer of the dove of the Holy Spirit,
and in one sample taken from the gilding of the rays in the coloured halo around
the dove. Four additional samples from the rays showed wood ash glass particles in
the gilding mordant. The introduction of powdered glass to accelerate the drying
of a mordant for gilding was a common studio practice,77 but its presence in a paint
containing well-drying lead white is more sporadic78 and suggests that the glass
may have been employed for a different purpose, for instance to modify the paint’s
optical properties and/or texture.
It should be noted here that establishing a clear difference between the wood-ash
and the wood-ash-with-some-fern-ash glass types is not straightforward because
of both the slight difference in the CaO/K2O values for these glass types and the
uncertainty associated with the edx spot measurement. Likewise, the leaching of
alkaline cations (e.g. sodium and potassium) from the glass particles in the course
of their ageing process can artifi cially increase the value of the CaO/K2O ratio.
Nevertheless, the analyses carried out on all the samples taken in 19500–51 from
the dove showed the same tendency: the CaO/K2O ratio was below one in nearly
all measured particles. The remarkably large size of the particles in the dove and
in one mordant sample was also taken into consideration. Indeed, the size of most
glass particles ranges between 10 and 600 µm, which is exceptional as the powdered
glass in the Eyckian paintings is usually fi nely ground with particle sizes typically
ranging between 5 and 15 µm.79
These results might suggest that the dove and some of the rays would have been
added in one single painting stage. But given that this type of glass can be found
in fi fteenth-century painting (including Jan van Eyck’s Arnolfini Portrait), the
dove cannot be dismissed as a non-Eyckian intervention based on purely material
grounds.80 Furthermore, taking into consideration that a dove of the Holy Spirit
was mentioned in 1458 as part of the tableau vivant representing the scene of the
Adoration of the Lamb presented at the Joyous Entry of Philip the Good into the
city of Ghent, it may be argued that the dove was present in the painting at that
moment.81 Likewise, as suggested in other chapters, the presence of another version
of the dove, hidden by the currently visible version, cannot be excluded.82
Cross-sections prepared from three dove samples taken in the 19500–51 analysis
campaign were reanalysed. The samples do not show a consistent stratigraphy;
they present two or three white layers that seem to belong to the same intervention
as there is no clear separation between the strata. In contrast, the white layers are
clearly separated from an underlying blue paint system that consists of two layers
of ultramarine applied over several layers of azurite. The sample taken from the
135
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Pb-L Pb-M
3.18a 3.18b 3.18c
wing (fi g. 3.18a) actually shows a thin fl uorescent intermediate layer between the
fi rst white dove layer and the ultramarine blue layers (fi g. 3.18f). This fi nding
suggests that they were not painted during the same stage. The results of the
ma-xrpd scanning of this area (fi g. 3.18d–e) support this interpretation. Based on
the various hc/(hc+c) calculated ratios, it is evident that the lead white used to
paint the dove (hc/(hc+c) = 600%) is different from that used for the surrounding
and underlying blue layers (hc/(hc+c) = 900%). It would seem therefore that the
dove was separately painted on top of a fully fi nished blue sky, which shows a paint
build-up similar to that of the rest of the sky on the Adoration of the Lamb. The fact
that the dove consists of two or three layers belonging to the same intervention
does not rule out the possibility that the current dove hides an earlier version not
136
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
The high-lime and low-alkali (hlla) glass with a CaO/K2O ratio between 5 and
8 was systematically detected in red glazes found in the fi rst overpaints (e.g. Joos
Vijd) and in many varnishes, located just below or on top of the fi rst overpaint,
indicating that they are very likely associated with the fi rst overpaint campaign.
The sodium oxide content was typically around 1.5–2 w%, and in some cases
higher (3.8–5 w%). It is very likely that this value is overestimated, since the
sodium might be partially derived from ultramarine. This type of calco-potassic
glass only began to be commonly used in the Netherlands at the end of the
fi fteenth century,83 and, according to Spring, it is very unlikely to have been used
before 14700.84 The identifi cation of the hlla-type of glass particles allowed us to
propose the 14700 date as a provisory terminus post quem for this early interference
with the polyptych.
Soda-ash glass
The most abundant use of glass particles was detected in the sixteenth-century
overpaint, which has been discussed in detail in the earlier book on the fi rst phase
of the Ghent Altarpiece treatment.85 The glass particles found at this level presented
a relatively consistent composition showing a high sodium oxide content (between
8 and 13.5 w%) and low concentrations of potassium, magnesium and phosphorus
oxides (2–3.3 w%, 1.8–2.7 w% and 0–1.1 w%, respectively), indicating the use
of soda-ash Mediterranean glass. This type of glass is usually detected in Italian
paintings and also in some German paintings dating from the end of the fi fteenth
century.86 Soda-ash glass was imported from Italy until the beginning of the
sixteenth century, when the fi rst Italian glassmakers arrived in the Netherlands
and started to produce it locally under the name of façon-de-Venise glass.87
This rather easily distinguishable type of glass turned out to be a very specifi c
indicator for the major sixteenth-century overpainting campaign, that was
carried out before 1557–58, the date of Michiel Coxcie’s copy of the polyptych,
in which these overpaints are reproduced. In this level, the powdered glass was
used in even larger quantities than in the older levels. It is interesting to add that
the earliest Netherlandish painting in which this type of glass particle has been
found so far is Colijn de Coter’s St John the Baptist and Six Apostles (Bayerische
Staatsgemäldesammlungen, Alte Pinakothek, Munich, inv. waf 163), presumed to
have been painted around 1510.88
137
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Zinc vitriol
As long ago as 1849, Mary Merrifi eld wrote: ‘There is a good reason to suppose that
white copperas (sulphate of zinc), which is mentioned as a dryer by Flemish and
German writers of the fi fteenth century, was the dryer of van Eyck. We owe this
discovery to the research of Mr. Eastlake.’89 This hypothesis was solely based on the
study of medieval and Renaissance technical treatises, without any material proof.
It took more than 1600 years to confi rm Eastlake’s and Merrifi eld’s assumption using
scientifi c analysis. The fi rst time the presence of zinc in the Ghent Altarpiece was
mentioned in an analytical report was in 20010.90 In the context of a preliminary
study prior to the fi rst restoration campaign, in-situ xrf spot analysis detected
an ‘intriguing’ presence of zinc in the red draperies of the central panel. It was
intriguing because the area was apparently not overpainted and, at that time, zinc
was commonly taken to be an indicator of modern zinc pigments. In the following
years, the re-analysis of two other Van Eyck paintings revealed the presence of
zinc in the original paint layers, including a violet paint layer from the Arnolfini
Portrait (The National Gallery, London, inv. ng186) and red and black layers in the
Portrait of Margaret van Eyck (Groeningemuseum, Bruges, inv. 0000 0000.gro00162.i).91
Besides the paintings by Van Eyck, zinc was found in many mid-fi fteenth-century
Cologne paintings.92 This research, together with Marika Spring’s recent article
summarizing current knowledge on the use of white vitriol as an additive for oil
painting,93 supports Eastlake’s assumption that white copperas was commonly used
by Netherlandish and German painters. Similarly, traces of zinc have been mostly
found in the red glazes and black paint layers in the Ghent Altarpiece panels. In
addition, zinc was found in many brownish shadows in ochre areas, in certain green
vegetation zones, in numerous dark brown colours such as the wooden ceiling
beams in the exterior upper panels, and in bronze (or brass)-like objects, such as
the chalice in front of the lamb, the column of the fountain in the Adoration of the
Lamb and the lavabo in the Interior View. In the aforementioned cases, the presence
of zinc was mainly demonstrated during the systematic ma-xrf screening of the
panels. Furthermore, zinc was also detected and localized in the paint stratigraphy
of the cross-sections by sem-edx.94 Even though the chemical identity of the
zinc compounds present in paint layers was quite diffi cult to determine, some
interesting results have already been obtained by sr-xanes, which confi rms the
presence of zinc sulphate, and by atr-ftir, which makes it possible to detect the
zinc carboxylates and oxalates.
In summary the current results suggest that the use of zinc vitriol as an additive
was a common practice in the two fi rst stages of execution. Interestingly, the
elemental zinc distribution observed in the Eyckian paint layers in the Pilgrims and
the Hermits panels, or the group of popes, the fountain and the vegetation above
the horizon line in the Adoration of the Lamb, is richer in this element than other
parts, such as the vegetation behind the group of confessors (fi g. 3.19a–d). The
later Netherlandish painters who altered the painting during the sixteenth century
also used zinc driers, as it was detected in the old overpaints. Certain areas, such as
the draperies of Elisabeth Borluut (fi g. 3.19e–f) or the Cumaean Sibyl, scanned with
ma-xrf during the removal of these overpaints, suggest that these layers are richer
in zinc than the underlying Eyckian layers.
138
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
original
Fig. 3.19a–f. ma-xrf distribution elemental maps for zinc (Zn-K) in the
Knights of Christ (a), Adoration of the Lamb (b), Hermits (c) and Pilgrims (d)
panels. The most intense Zn signals are found on the right side of the
Adoration (b) as well as in the Hermits (c) and Pilgrims (d) panels, in
overpaint particular for the cypress trees behind the horizon line. Strong Zn-K
signals observed in luminous and irregular spots correspond to paint losses
in-fi lled with a modern zinc white (ZnO)-containing material. Detail of
pink drapery of Elisabeth Borluut halfway through overpaint removal (e)
and the corresponding ma-xrf Zn-K distribution map (f) showing a more
intense Zn signal in the overpaint zone.
3.19e 3.19f
When and how was the drier introduced into the paint? Was it already incorporated
into the binder or was it added to the pigment before the paint was made? Was it
added on the grinding slab during paint production or was it mixed on the palette?95
In the case of red lakes, a careful examination of the macro- and micro-analytical
results suggests that the additive is related to the lake pigment rather than to
the binder. For instance, in the ma-xrf Zn-K maps, the most intense signals
are observed in the lake-containing red and pink clothes of the fi gures in the
Adoration of the Lamb, in St Anthony’s purple drapery in the Hermits (fi g. 3.15) and
in the drapery of St Christopher in the Pilgrims.96 The signal is likewise intense
when multiple, superimposed lake-containing layers are present. An example of
the latter was evidenced by comparing ma-xrf results with three samples taken
from St Christopher’s drapery, in light, half-shadow and shadow zones in which
respectively two, three and four red lake-containing layers were observed. In the
light areas of the modelling, the lake is mixed with lead white and the intensity
of zinc seems much lower, as shown by the ma-xrf Zn-K map. Conversely, the
intensity of the ma-xrf zinc signal in the shadow zones is high.
ma-xrf maps also revealed a series of smooth-edged losses in St Anthony’s dark
purple mantle. As illustrated by the Pb-L and Zn-K ma-xrf map (fi g. 3.200b–c),
139
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Pb-L Zn-K
these losses occurred at an early stage in the creative process. Indeed, they seem
to have been infi lled with a lead-containing material and retouched with a zinc-
containing red lake. The reason for these losses could be a technological problem,
for example an excess of drying agent(s), resulting in strong drying cracks which are
still observable in this area (fi g. 3.200d). The atr-ftir analysis of one sample taken
in this area contributed to a better insight into the latter phenomenon. By this
method, zinc and lead carboxylates were indeed found in all red lake-containing
glazes. Their presence suggests the use of zinc- and lead-based driers that reacted
with the oil binder.
A zinc drier was also typically used in the brownish areas presenting a transparent
glaze layer (fi g. 3.21) that consists of an organic brown colourant. The latter has not
yet been identifi ed but is assumed to be either a bituminous coal-based pigment
– bistre, asphalt, aloe, pitch or similar materials retarding the drying of oil. The
zinc additive was consistently introduced in such brown glazes, for instance in
the shadowing of the earthy paths on which the fi gures of the Knights, Hermits
and Pilgrims panels stand. Interestingly, it was also observed in more opaque
dark brown and green mixtures containing pigments such as verdigris or brown
ochre which, unlike the colourants, advance the hardening of the oil binder. This
brownish transparent paint, seemingly prepared with a substantial concentration
of zinc vitriol, was observed under the binocular microscope as well as in samples
taken from the leaves, cypresses and palm trees painted above the horizon line in
the Pilgrims and the central Adoration of the Lamb panels. It seems that they were
not planned in the fi rst stage, but were possibly added subsequently, as discussed
in the previous chapter.97
Likewise, the fountain in the middle of the Adoration of the Lamb is considered to
be a compositional modifi cation introduced in the second stage. As suggested by
the Zn-K and Fe-K ma-xrf distribution maps in this area, the central brownish
column, rising from the dark green water of the fountain, is painted with brownish
140
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Zn-K Fe-K
ochre-containing mixtures with quite high amounts of zinc additive (fi g. 3.21a–c).
A sample taken from the area where the column is below the water level shows
two thin, brown ochre-containing layers on top of the stratigraphy. Two other
strata are visible under the brown layers: a bluish-green one applied over a black
one (fi g. 3.22a–b). Taking into account the hypothesis proposed in chapter 2, this
black layer possibly corresponds to a dark hole in the meadow (a ‘spring’ painted in
the fi rst stage of the creative process).98
Surprisingly, sem-edx mapping suggests that zinc is not only present in the
brown upper layers, but dispersed in all the strata in very low amounts (fi g. 3.22c).
atr-ftir microscopy indicated that this element is mostly present in the form of
zinc carboxylates. Interestingly, copper and lead soaps, as well as copper oxalate and
formate, were also detected throughout the sample’s strata (fi g. 3.22f), suggesting
that the binder has thoroughly reacted with the pigments and driers employed to
produce these paint layers.
Besides zinc, sem-edx analysis of the top brown layers indicated the presence of
some lead, copper, calcium and phosphorus (fi g. 3.22c), the latter being found in
the form of phosphates as revealed by atr-ftir. Accordingly, these results suggest
the presence of calcium phosphates in these layers, but further research is required
to defi ne their precise chemical identity. Such simultaneous addition is reminiscent
of the recipes to accelerate drying mentioned in the Strasbourg Manuscript.99 The
zinc vitriol and the calcined bones can be added on the grinding slab (‘… to all
above mentioned colours you must add a little calcined bone or a little zinc vitriol
– as much as a bean – in order to make the colour dry well’) or in the preparation
of the oil called oleum preciosum, described as ‘very quick drying and [it] makes all
colours lovely, clear and bright’.100 Given that such a combination of compounds has
also been detected in other Ghent Altarpiece samples, we consider it very likely that
Van Eyck used one or both methods described by the German anonymous author.
As shown by the paint cross-sections analysed so far, zinc vitriol and powdered
glass were not used together in the same layer. It was also very rare to fi nd them
in a sequence of layers that can be attributed to the same painting intervention.101
However, both are frequently found in combination with other compounds such as
verdigris, calcined bones or lead-based pigments which, in certain conditions of the
141
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
50
0 μm 50
0 μm Zn Cu Pb Ca P Electron
Moolooite
1800
00 1700
00 1600
00 1500
00 1400
00 1300
00 1200
00 1100
00 1000
000 9 00
Zn, Cu & Pb carboxylates Cu oxalates
Wavenumber (cm-1)
3.22d 3.22e 3.22f
Fig. 3.22a–f. Adoration: paint cross-section from the column below the water surface (fi g. 3.21) (10/83) under vis
light (a) and back-scattered electron sem image (b). sem-edx spatial distributions (c) of copper (Cu-K, green),
zinc (Zn-K, red), phosphorus (P-K, dark blue), lead (Pb-M, yellow) and calcium (Ca-K, cyan). atr-ftir analysis
of the cross-section enabled us to obtain spatial distribution maps of metal carboxylates (d) and copper oxalates (e).
An ftir spectrum (red), extracted from the black layer (dotted white square, e), shows characteristic copper oxalate
(moolooite) vibrations. Moolooite reference spectrum (blue) is included for comparison (f).
preparation of the paint mixture, can promote the formation of metal carboxylates,
themselves enhancing the drying properties of oil.102 These metallo-organic driers
are nowadays classifi ed into primary (or active fast) and secondary (or auxiliary),
with lower catalytic activity, and it has been demonstrated that their combination
allows a fast and ‘thorough’ drying without causing certain defects such as drying
cracks or ‘wrinkling’.103 The examination of the Ghent samples makes us think
that the Van Eyck brothers empirically mastered this subtle association and dosage
of additives (Pb, Zn, Cu, Ca-P), producing a liquid paint with suitable handling
properties which could quickly convert into a good-quality paint fi lm.
142
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
Conclusions
The scientifi c investigation of the Ghent Altarpiece panels restored so far has allowed
us to establish an already quite robust, general overview of the Eyckian materials
and painting techniques.
Some of the results allow us to complement previous material studies of the
polyptych, for instance with respect to the identifi cation of a series of new pigments
and additives (coal black, bistre, kermes lake, orpiment, bone white, white vitriol
and colourless powdered glass), which enlarge the list of previously found paint
materials in the original layers.
However, the ongoing scientifi c research has also allowed us to unravel novel details
which should make it possible to refi ne our comprehension of the Ghent Altarpiece’s
creation process. For example, the comparative study of the priming layers applied
in different panels of the altarpiece has demonstrated material similarities between
certain panels and enabled us to categorize them into three main groups. Similarly,
the quantitative study of the glass particles in various paint levels permitted us
to propose a hypothesis correlating the material composition of the different glass
types found in the altarpiece and the chronology of execution of the paint layers
containing them. Finally, the detection of various additives supposed to accelerate
the drying of the oil binder, such as zinc sulphate, calcium phosphate, lead- and
copper-based pigments, paves the way towards a better understanding of how these
associations and minute dosages allowed this huge work of art to be painted.
Notes
1 Coremans 1953. 8 Coremans 1953; Brinkman et al. 1984/85;
2 Coremans 1953; Van Asperen de Boer 006; Clark 20011; Vandivere 20013;
Faries 200
1 979; Brinkman et al. 1984/85, 1990 0; Steyaert, Postec 200200.
Billinge et al. 1997a; Campbell 1998a; 9 During the 19500–51 campaign, scientists
Gifford 1999; Deneffe 20011; Gifford, suggested that the study of coccoliths
Metzger, Delaney 20013; Gifford et al. present in the altarpiece’s chalk ground
20017; Spring, Morrison 20017; Dunkerton, could help to differentiate studio
Morrison, Roy 20017. practice, but that aspect was not further
3 Spring 20012; Spring 20017; Spring, explored: Dubois, Deneffe 20019. In the
Morrison 20017. next study of these old samples, the
coccoliths revealed by sem-edx were
4 Van der Snickt et al. 20017; Sanyova et al. identifi ed as originating in Western
200200. Europe: Brinkman et al. 1984/85.
5 The methods mentioned are described in 10 Sample from the prophet Zechariah’s
the previous publication devoted to the pink drapery was analysed at the
fi rst phase of restoration, Sanyova et al. European Synchrotron Radiation Facility
200200. (esrf), Grenoble, as part of the hg41
6 Gifford 1999; Gifford, Metzger, project; Beamtime with sr-xrf, sr-
Delaney 20013; Billinge, Campbell 1995; xanes and sr-xrpd at id 21 and id 13.
Campbell, Foister, Roy 1997. 11 The painting of the Just Judges was
7 Jones, Augustyniak, Dubois 200200. stolen, but its frame was not and it
143
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
remained with the other frames of the 29 Plahter 1992; Nadolny 200 001, pp. 426
wing panels. and 445.
12 Neven 20016, pp. 121 and 1600. 30 Measurements with sr-µ-xrpd were
13 These techniques are described in detail carried out at the id13 (esrf, Grenoble)
on the basis of the irr images in chapter in March 20014.
2 of this volume. 31 The addition of highly alkaline
14 https://www.researchgate.net/ quicklime to the leaded oil is
publication/248517293_Chlorine_in_ recommended by Archerius in 1398 for
coal_A_review (accessed 12 November the preparation of a heat-bodied oil for
200200). tempering pigments: ‘Si vous voulez
15 Coremans 1953; Brinkman et al. 1984/85; appareiller oile pour destemper toutes
Campbell 1988; Campbell, Foister, Roy manieres de couleurs. Prenes chaux
1 997; Stroo 200
009, pp. 122–23, 153–55, vive avec autant de ceruse cofi rne est
2700–71, 417–19. de loile, puiz metez au soleil et ne le
movez jusques a ung moyt ou plus tar
16 Campbell, Foister, Roy 1997a,b, pp. quant plus y sera, et mieulx vaudra, puis
22, 71; Stege et al. 200
007, pp. 74–75; le colez et gardez tres bien loile, et de
Nadolny 200 008; Faries 20011; Spring 20
012; celle oille gardee et ainsi preparee, povez
Spring 20017. destremper toutes couleurs ensemble et
17 Ferreira 200
006; Gifford, Metzger, Delaney chacun par soy.’ In De diversis coloribus
20013; Vandivere 20013, p.75; Van Baum no. 319; Merrifi eld 1849, p. 3003.
et al. 20014, p. 49. 32 Keller 1973; Bielman 1993.
18 Nadolny 200 008. 33 Cotte et al. 20016.
19 Chalk was detected in several samples
with sem-edx thanks to the presence of 34 A similar intermediate layer was
fragments of coccoliths. observed in the National Gallery of Art
Annunciation.
20 No reliable sample from the Eve panel
was available to confi rm whether or not 35 Campbell, Foister, Roy 1997; Campbell
1 998a.
it also belongs to this same group.
21 Van Mander (16004); see Vandivere 20013, 36 Coremans 1953; Brinkman et al. 1984/85;
p. 76. Brinkman et al. 1988/89.
22 It is interesting to note that the priming 37 The copper green pigment identifi ed by
in the upper part of the Archangel panel Thissen as malachite in all green samples
with the prophet Zechariah is darker taken in 19500–51 is in fact verdigris.
than that used in the part with the 38 It should be noted that the hydrocerussite
Archangel Gabriel. and cerussite present in the priming
23 The bone white-containing layer was not layers could also have been formed by the
detected in all samples, which makes it remineralization of the lead soaps from
uncertain that it was applied over the a leaded oil, as described by Cotte et al.
entire ground. Klaassen et al. 20021. 20016.
24 Plahter 200003. 39 Gonzalez et al. 20017.
25 https://www.metmuseum.org/blogs/ 40 This attempt to recognize various
collection-insights/20019/jan-van-eyck- qualities of lead white, using sr-xrpd
frames-imaging-xray-xrf (accessed 12 on several microsamples from the Ghent
November 200200). Altarpiece panels, showed the potential
26 Stroo 200009, pp. 2700–71. of such an approach. The long-term
lead white evolution is the topic of an
27 Bone white is also mentioned as an ongoing research led by the University of
extender of the ground on a dry wall as Antwerp’s axes group.
an alternative to gypsum or chalk in the
Montpellier Liber diversarum artium (Clark 41 Gonzalez et al. 20017.
20011, p. 143). Cennini mentions using the 42 Vandivere, Clarke 20012.
calcined bone in chapters 5–7; Cennini 43 For a further discussion of these
(ed. Thompson) 19600, pp. 4–5. interventions, see chapter 2, sections ‘The
28 Neven 20016, pp. 121, 123 recipes 66, 68, sky and the horizon’ and ‘Architecture’
pp. 161–162, 184, 189. and chapter 4.
144
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
44 Van Asperen de Boer 1974; Spring, 60 Borrodaile 1966, p. 49; Merrifi eld 1967,
Morrison 20017. pp. 24, 27.
45 Coremans 1953; Campbell, Foister, Roy 61 Von Baum et al. 20014, p. 1400.
1 997; Campbell 1998a. 62 Roldan et al. 20014; Baumer et al. 20019.
46 Spring, Morrison 20017; Spring 20017; Bistre is so fi nely dispersed in the washes
Gifford et al. 20017. obtained through the reconstruction
47 Ibid. of old recipes that it forms a thin
48 The same biomarker, the indigotine discontinuous barely visible layer and it
molecule, is commonly used for the would be quite hard to fi nd an area large
identifi cation of both the European enough to be suitable for sampling.
and the Asian plants, which cannot be 63 The samples containing an original red
differentiated on this basis. glaze were available in very low amounts
49 Kirby 200 006. and were nearly always contaminated
by later interventions, as shown by the
50 Borrodaile, Borrodaile 1966, p. 61. optical microscope observation of the
51 Von Baum et al. 20014, pp. 137 and 336. cross-sections under uv light.
52 Eastaugh et al. 200004, p. 385; Buse et al. 64 This was based on the detection of the
20019. kermesic acid and laccaic acid A and E.
53 Scott, Taniguchi, Koseto 200 001; Buse et 65 Sanyova et al. 200200.
al. 20019.
66 Kirby, Spring, Higgitt 200 005.
54 Borrodaile, Borrodaile 1966; Neven 20016.
67 The analysis was carried out and
55 Scott, Taniguchi, Koseto 200 001; Buse et interpreted by Dr Stepanka Kuckova
al. 20019. at the University of Chemistry and
56 The mechanism of formation of metal Technology, Prague.
oxalates in Southern Netherlandish oil 68 Kirby 200008; Kirby et al. 20014.
paintings, including the Ghent Altarpiece
panels, is currently under investigation 69 See note 12.
in the MetOx project (belspo br/165/a6/ 70 Lutzenberger, Stege, Tilenschi 20010;
metox) coordinated by kik-irpa. Spring 20012; Spring 20017; Spring,
57 The use of zinc-containing compounds Morrison 20017; Melo et al. 200200.
is discussed in the section related to the 71 Campbell, Foister, Roy 1997a; Campbell
additives. 1 998a; see also chapter 2 in this volume.
58 During the study in 19500–51, all copper- 72 Pacheco’s and Paduan manuscripts are
based green pigments were described as cited by Merrifi eld (1849) 1967.
malachite and all transparent green glazes pp. ccxxxix–ccxlii.
as copper resinates (Coremans 1953,
pp. 71–73). The updated analyses in 73 Lutzenberger, Stege, Tilenschi 20010; Spring
1 988/89 have confi rmed that copper
20012; Spring 20017; Melo et al. 200200.
resinate is present in the transparent 74 Lutzenberger, Stege, Tilenschi 20010;
layers, while malachite was not found. Spring 20012.
Consequently, the green pigment 75 Spring 20012; Spring 20017.
was described as “copper based” only
76 The glass-containing red overpaint
(Brinkman et al. 1988/89).
detected in this sample from the
59 Both samples were analysed with Archangel’s wing (13/37) was described
pyrolysis-gas chromatography–mass as a second overpaint in the fi rst phase
spectrometry (Py-gcms) after their of treatment: Sanyova et al. 200200.
methylation with (tetramethyl) The updated comparative study of the
ammonium hydroxide (tmah). glass powder compositions in the Ghent
The positive detection of bistre was Altarpiece samples suggests that this glaze
based on the presence of methoxyphenol was applied before the overall sixteenth-
markers, deriving from pyrolyzed century overpaint.
lignin, the guaiacol and syringol
derivates (2-methoxyphenol and 77 Lutzenberger, Stege, Tilenschi 20010;
1 ,3-dimethoxyphenol respectively) as Spring 20012; Spring 20017.
well as polycyclic aromatic hydrocarbons 78 Lutzenberger, Stege, Tilenschi 20010 has
(pahs). Baumer et al. 20019. found the soda ash glass powder in the
145
3. THE CHALLENGES OF A COMPLEX STRATIGRAPHY
fl esh tone of a painting by Colijn de Coter minor concentrations that might easily
dated before 1510. This type corresponds fall below the minimum detection
to the cristallo type of soda glass. limit of the analytical method used,
79 A similar case was observed in the e.g. sem-edx. The detection limit of
Tiefenbronn Altarpiece by Lukas Moser sem-edx is typically in the order of
(dated 1432) with a particle size up to 00 parts per million (ppm) while the
200
700 µm. However, the composition of ma-xrf instrument used for the Ghent
this glass was different from that found Altarpiece was able to visualize zinc
in the dove. Two types of glass were concentrations at or just below 100 ppm;
found in the gilding mordant in Moser’s Alfeld et al. 20011.
altarpiece, one sodium-rich, Venetian 95 Spring mentioned (20017) that zinc vitriol
soda ash glass (vitrum blanchum) and one could have been used for the preparation
with quite high content of sodium and of a special quick-drying oil, oleum
potassium, considered as an intermediate preciosum, or added as a powder on the
mixed alkali type; see Lutzenberger, grinding slab according to the Strasbourg
Stege, Tilenschi 20010. Manuscript. According to the Tegernsee
80 Spring 20012. Manuscript, it could also have been used
81 Dubois, Sanyova, Vanwijnsberghe 20017. as an ingredient in the manufacturing
of a high-quality red lake, having better
82 The authenticity of the dove is discussed drying properties.
in the previous chapter 2, section ‘The
dove, the halo and the rays’ and in 96 Care should be taken when an attempt
chapter 4. is made to give semiquantitative
83 Janssens et al. 1998; Caen et al. 200006. signifi cance to ma-xrf plots, as various
unknown parameters can infl uence the
84 Spring 20017: hlla-type glass was resulting pixel intensities, such as the
identifi ed in ten out of nineteen potential accumulation of Zn-based
Netherlandish paintings in the National layers, their possible fragmented lateral
Gallery, London, all dated after 14700. distribution and heterogeneous paint
85 Sanyova et al. 200200. layer thicknesses.
86 Lutzenberger, Stege, Tilenschi 20010; 97 Chapter 2, section ‘Trees’.
Spring 20012; Spring 20017. 98 Chapter 2, section ‘Meadow and fountain’
87 Janssens et al. 1998. and fi g. 2.14.
88 Lutzenberger, Stege, Tilenschi 20010. 99 Borrodaile, Borrodaile 1966, pp. 55,
89 Merrifi eld 1849, p. ccxlii. 61, 65.
90 The project Lasting Support (1 April 100 Ibid., p. 55.
20010–300 June 20011) was supported by 101 This was the case of the red mantle of a
The Getty Foundation. The analyses were bald philosopher on the lower left in the
carried out within the charisma project Adoration panel, where two red layers
by molab. with lake from wool shearings containing
91 Spring, Morrison 20017. zinc vitriol (10/73, layers 5, 6) were
92 Stege, Tilenschi, Sanyova 20012; Spring covered with a third red glaze (10/73,
et al. 20012; Von Baum et al. 20014, layer 7) containing wood ash-type glass
pp. 133–1400. particles.
93 Spring 20017. 102 Cotte et al. 20016.
94 As an additive, it is often present in 103 Bielman 1993.
146
4
One Painter or Several? A Stylistic Study
Technical and scientifi c research has enabled us to distinguish three stages in the
development of the lower register of the altarpiece interior. Are they evidence of a
single painter adjusting his composition as the creative process unfolded?1 Are they
the work of Hubert and then of Jan, as the quatrain states? Or may we attribute
the fi rst stage to Hubert, the second to Jan and the third to assistants or even
to a later painter? Neither the technical scientifi c investigation nor the chemical
analysis are suffi cient in themselves to attribute these successive interventions with
certainty. A stylistic approach is thus still required in any attempt to establish
whether different hands might be in evidence. The only known quantity in this
equation with multiple variables is Jan van Eyck.
Eyckian interventions
Jan
Besides the Ghent Altarpiece, Jan van Eyck’s oeuvre comprises nine paintings that
are signed and dated between 1432 and 1439,2 to which we may add a few widely
accepted attributions.3 These works are used here by way of comparison. Jan van
Eyck’s keen sense of observation and his extraordinary ability to represent what
he observed enabled him to translate the many nuances of colour produced by the
refl ection of light on materials of every kind, thus making the world he painted
tangible.4 In the Adoration of the Lamb, the painter conveyed the many shades
observed in the metal column of the fountain, in the same way as in the Virgin
at the Fountain,5 through more or less warm tones but also by touches of varying
density – from an intense highlight rendered through a dense pictorial substance
to fi laments of diffuse refl ection.6 A dragon’s head in the fountain of the Adoration of
the Lamb closely resembles a similar detail of the chandelier in the Arnolfini Portrait
(fi g. 4.1a–b).7 On top of a brown silhouette, Jan suggested refl ections by applying
(facing page) a few yellow dots – tiny, but precisely distributed in accordance with the lighting
Adoration of the of the represented location.
Lamb: head of a The veins and cavities specifi c to the various materials of the stone and marble
prophet attributed basin in the Adoration of the Lamb are represented, as are two characteristic fragments
to Hubert van Eyck of a chipped stone (fi g. 4.2b). Similar details are also found in the Virgin of Chancellor
(see also fi g. 4.5). Rolin and the Virgin of Canon Van der Paele,8 right down to the chips in the stone,
149
4. ONE PAINTER OR SEVERAL?
4.1a 4.1b
Fig. 4.1a–b.
indicated by two simple touches of paint, one brown, the other white (fi g. 4.2a Refl ections on
metal. Jan van
and c). The waves created by the water of the fountain as it falls into the basin in
Eyck: detail of
the Adoration are painted with nuanced strokes (fi g. 4.3a). The water splashes back the fountain,
up at the point of impact to form suspended droplets, grouped together and then Adoration of the
bursting, in precisely the same way as in the Virgin at the Fountain (fi g. 4.3b). The Lamb (a); detail
rendering of tufts of grass and certain small fl owers in the second-stage meadow of of the chandelier,
the Adoration is likewise comparable with what Jan painted at the feet of the Virgin Arnolfini Portrait,
at the Fountain and in the fl owerbeds of the Virgin of Chancellor Rolin.9 Through 1 434, London,
the very shape of the touches of paint and their precise distribution, through the The National
orientation of his brush, and the gesture he made when applying the paint, he not Gallery, inv.
only gave shape to the plants he painted, but also captured what distinguishes lily- ng186 (b).
of-the-valley, say, from buttercup, no matter how small the fl ecks of colour. This
extraordinary ability to translate such refi ned observations into painting through
swift yet never accidental brushstrokes is found throughout Jan van Eyck’s work.10
His touches combine precision with speed (seemingly contradictory characteristics),
enabling him to vary his effects.
In his hands and faces, Jan did not simply paint the colour of the skin: he
incorporated all the nuances created by veins, blood vessels and imperfections of
the epidermis. The fl esh tones of the prelates in the Adoration (fi g. 4.4a) resemble
those of St Donatian in the Virgin of Canon Van der Paele (fi g. 4.4b) and the donor
in the Virgin of Chancellor Rolin (fi g. 4.4c). Stubble is indicated by bluish areas of
skin, patches of red in the epidermis are fi nely distributed, brown lines suggest
wrinkles that furrow the fl esh. Two fi ngers of the left hand of the philosopher in
the blue cloak are lit from behind (fi g. 4.14b), just like Arnolfi ni’s right thumb
(fi g. 4.14a)
All these characteristics associated with Jan van Eyck are found in the second
stage.
150
4. ONE PAINTER OR SEVERAL?
151
4. ONE PAINTER OR SEVERAL?
4.3a 4.3b
Fig. 4.3a–b.
Splashing water
apostle in the Adoration had already been painted when Jan intervened and that droplets and waves.
he decided to retain him.15 A sample16 taken from the spot where the apostle’s Jan van Eyck: detail
robe overlaps the fi rst-stage meadow tells us that this meadow had been painted, of the fountain,
right up to the glaze, before the fi gure was added. The fi gure and the fi rst meadow Adoration of the Lamb
may thus belong to the same stage. Bearing in mind the information provided by (a); detail of the
the quatrain, we may indeed conclude that this apostle could be Hubert’s work, fountain, Virgin at
together with the fi rst meadow and the spring, now concealed.17 the Fountain, 1439,
The fi rst apostle (fi g. 4.5b) is not the only fi gure that may be attributed to Antwerp, Royal
Museum of Fine
Hubert: the one kneeling to his right in the front row and another a little further
Arts, inv. 411 (b).
down (fi g. 4.5a) belong to the same category of fi gures that we will refer to as ‘less
realistic’, as do some of the faces of the fi rst prophets to the left of the fountain.18
The eye zone (fi g. 4.5a) shows the same characteristics as those observed in the fi rst
apostle (fi g. 4.5b). In the prophet with the red hat (fi g. 4.5c), we fi nd the same
dark line between the upper eyelid and eyeball, terminating in a curl where the
wrinkles begin. The face of the fi gure in the blue cap, partly hidden by the green
drapery, also displays features that seem far removed from Jan’s art (fi g. 4.5d). In
the fi gures painted by Hubert, wrinkles are schematically indicated rather than that
they convey a precise topography of the creases in the skin. They lack relief, as does
the volume of the cheek. The brown lines in the hermit and pilgrim painted by Jan
are combined with light touches that perfectly suggest the irregularities in the folds
of the fl esh (fi g. 4.6a–b). Moreover, the difference in colour in the skin between the
wrinkles at the corner of the eye and those in the cheeks lends volume to the faces.
There are likewise disparities between the meadow next to the horses in the
Knights of Christ and that of the Adoration of the Lamb. In the Knights panel, the
grass is grouped into tufts that form a stylized motif, unlike the airy grass of the
152
4. ONE PAINTER OR SEVERAL?
153
4. ONE PAINTER OR SEVERAL?
4.5a 4.5b
4.5c 4.5d
Fig. 4.5a–d. Adoration of the Lamb, heads attributed to Hubert van Eyck.
154
4. ONE PAINTER OR SEVERAL?
4.6a 4.6b
4.6c 4.6d
Fig. 4.6a–d. Hermits, head attributed to Jan van Eyck (a); Pilgrims, head attributed to Jan van Eyck (b);
Adoration of the Lamb, heads attributed to Hubert and reworked by Jan van Eyck (c and d).
155
4. ONE PAINTER OR SEVERAL?
central panel. Analysis of a cross-section19 shows that a second meadow has not been
painted here over a fi rst,20 as is the case in the central panel, and that only two layers
of green are present above a brown layer. This meadow zone in the Knights may thus
be attributed to Hubert.
The differences are not limited to the richness of nuances and variety in the
brushwork: Jan, who painted the hermit and the pilgrim (fi g. 4.6a–b), was clearly
more concerned with observing and recording reality and perhaps also better able to
translate these refi ned observations into paint. Both artists used the same technical
means, but there is a world of difference in their respective visual languages.
Based on the characteristics outlined above, the painted faces in the lower
register may be roughly divided into two groups – one clearly more faithful to
anatomical reality than the other. The less realistic faces are much fewer in number
(fi g. 4.7). Some authors mention the possible participation of Jan’s or possibly
Hubert’s assistants and hence of two workshops.21 It seems unlikely, however, that
an assistant would have painted less realistic fi gures in such a central part of the
composition or that Jan would have worked so meticulously around an assistant’s
contribution. The two groups mentioned above are coherent and, where differences
exist within them, these were consciously introduced by each of the two brothers.
Hubert and Jan both sought to vary the fi gures’ faces, as well as taking account of
the direction of the light in their execution. The fi gures to the left of the fountain
are fully illuminated, while those on the right partially turn their backs to the
light, which results in differences of colour.22 Some of Hubert’s heads, lastly, were
probably reworked or retouched by Jan. The eye of one of the apostles, for example,
was given a more natural shape through the addition of a white highlight and a
few shadow lines (fi g. 4.6c). The face of one of the prophets, meanwhile, in which
traces of rosacea were not really integrated in the modelling, had the eyebrow and
wrinkles retouched (fi g. 4.6d).
All the less realistic heads are grouped in the centre of the composition, in two
specifi c zones on either side of the fountain, close to the edges of the fi rst lawn,
which had already been painted (fi g. 4.7). These are the very fi gures within each
group that were most likely to be painted fi rst in a fi fteenth-century panel. Once
the underdrawing had been completed, the logical elements to be painted fi rst
would have been large expanses such as the sky and the meadow, leaving a reserve
for other parts of the composition, including the groups of fi gures.23 These reserves
were then fi lled in a logical order, working from the background to the foreground
and superimposing fi gures further back in the group with the ones nearer the front,
so that the pictorial layer of each fi gure slightly overlaps that of any others behind
it.24 Figures on the edge of the reserved space slightly overlap the edges of the fi rst
meadow.
Certain parts of a painting – especially in an ensemble as large as the Ghent Altarpiece
– will have been completed while adjoining areas were still at the underdrawing stage
or had only been given their fi rst layers of paint. This is illustrated by unfi nished
156
4. ONE PAINTER OR SEVERAL?
works like the St Barbara in the Royal Museum of Fine Arts Antwerp.25 Hubert
appears to have followed this general pattern of working from the back to the
front. Having painted the sky, the architecture on the horizon, the meadow and the
spring, he started work on the large fi gures in the foreground. He began with those
nearest to the spring at the centre of the composition and progressed outwards. At
some point, his work was curtailed and subsequently resumed by Jan.26 Hubert’s
contribution seems to have halted after the execution of the fi gures in the vicinity
of the spring. The further away we move, the more we fi nd characteristics of Jan’s
hand: in most of the philosophers to the left of the fountain, in the apostles located
somewhat further away on the right, and in the fi gures of the churchmen.27
We have seen in the central area how the second-stage meadow skirts the fi rst-
stage altar, lamb and angels. Even if Hubert had laid down the basic foundation of
these fi gures, the fi nal result seems to be Jan’s. While there are no securely dated
Fig. 4.7. Adoration of works by Jan that include animals similar to the lamb, he often painted the fur
the Lamb, diagram: lining of garments with the same accurate rendering of specifi c texture that we
heads attributed to fi nd here in the lamb’s coat. The eyes, meanwhile, with their horizontal slit pupils
Hubert (hatched typical of prey animals (fi g. 1.8c),28 may be compared with the eyes of the dog in
in yellow) and
the Arnolfini Portrait or the eye of the collared parakeet in the Virgin of Canon Van
to Jan van Eyck
(hatched in red) and
der Paele29 (fi g. 4.19), each of them painted with an amazing attention to the specifi c
heads attributed morphology of each species.
to Hubert and The virgin martyrs and confessors groups can also be attributed to Jan. The faces
reworked by Jan van and hair of the female saints are very similar to those of St Catherine in the Dresden
Eyck (hatched in Triptych (fi g. 4.8), although the lighting is different.30 The fact that the second
yellow and red).
4.7
157
4. ONE PAINTER OR SEVERAL?
4.8a 4.8b
Fig. 4.8a–b.
Adoration of the Lamb,
meadow was laid down around these fi gures could mean that Hubert had already heads of virgin
painted a fi rst version, perhaps even to a very advanced stage, as suggested by a martyrs attributed
cross-section from the blue gown of one of the virgin martyrs, which comprises to Jan van Eyck
fi ve layers.31 (a); Jan van Eyck,
Our hypothesis that a second stage attributable to Jan van Eyck covers a fi rst Dresden Triptych,
stage previously executed by Hubert and already well advanced in its painting, detail of the head of
makes it highly likely that the preparatory layers too were done by Hubert.32 Some St Catherine, 1437,
authors have suggested that the preparatory drawing might, in theory, be his.33 Dresden, Staatliche
Kunstsammlungen
Others, studying the infrared images, have noted divergences in the underdrawing
Dresden, inv. 799 (b).
of the Ghent Altarpiece without concluding that it might have been done by a
painter other than Jan or otherwise raising the possibility that Jan’s assistants had
a hand in it.34
The style of Jan van Eyck’s drawing, as revealed by infrared refl ectography in
signed and dated works other than the Ghent Altarpiece, shows a high degree of
coherence. As several authors have pointed out,35 the main folds of the draperies are
delimited by more strongly marked strokes, and the volumes are shaped using a
more or less dense framework of fi ne, straight lines, set down parallel to the contour
of the forms they defi ne. These longer or shorter lines intersecting in deep shadows
may bend slightly, but are very rarely curved. The drawing of St Barbara provides
a good example.
The characteristics of the underdrawing observed in the Adoration of the Lamb
differ from this,36 most notably in the nervous strokes, often curved and crossing the
broad lines that demarcate the folds, creating a more disordered impression than
Jan’s drawing. This type of drawing is clearly visible in the pale-coloured robes of
the prophets, apostles and angels (fi gs. 4.9 and 4.10).37 These short lines, can also be
158
4. ONE PAINTER OR SEVERAL?
Non-eyckian interventions
159
4. ONE PAINTER OR SEVERAL?
4.10a 4.10b
Fig. 4.10a–b.
Adoration of the
Examples include many of the trees in the shutters and the central panel, where Lamb, prophet:
these additions were executed in a highly repetitive manner (fi gs. 2.55–2.58).40 The irr details.
painter in question applied them with a routine gesture antithetical to Jan’s art.41
The sky and mountains of the central panel, all based on ultramarine, were
painted at the same time, and this after the buidings were fi nished. The openings
of some of the turrets are fi lled with the ultramarine-based layer (fi g. 2.46b),
while others are not (fi g 2.36b). We also fi nd certain places along the horizon
where the ultramarine covers fi ne painted details, such as the ornamentation and
highlights of some of the turrets or the crest of a roof (fi gs. 2.36, 2.37 and 2.46).
Tiny details like this and highlights were generally the last things to be added
in early Netherlandish painting.42 The City View in the altarpiece exterior reveals
that all the architectural details standing out against the sky were painted after
the application of the fi nal blue layer.43 The sky in St Barbara was nearly fi nished,
even before the painting of the tower had begun.44 The ultramarine layer, fi nally,
makes the sky in the central panel milkier than the deep, subtly gradated skies
of the shutters. All these features cast doubt on the Eyckian authorship of this
ultramarine-based fi nal layer.
160
4. ONE PAINTER OR SEVERAL?
4.11a 4.11b
Fig. 4.11a–b.
Pilgrims,
St Christopher’s red A horizon line and snow-topped mountains that may be compared stylistically to
cloak: vis and ir the mountains in the Ghent Altarpiece (fi g. 4.12) are found in the Virgin of Chancellor
details. Rolin. Those in the Ghent Altarpiece are fl at, each painted with a single blue tone,
whereas the mountains in the Rolin Madonna have a much more rugged relief within,
with a wide variety of touches and nuances. The fortresses standing in the mountains
in the Knights of Christ and in the Rolin Madonna (fi g. 4.13) are both painted in
similar tones but to very different effect. Brilliantly distributed brushstrokes evoke
the impression in the latter of a citadel surrounded by a defensive wall that sparkles
in the morning sun.45 Each pink touch gives effective shape to a tower, a wall or
even the face of the rocky outcrop on which the building stands. The mountains in
the Knights, by contrast, are more like fl at stage scenery, the fortifi ed town is a two-
dimensional silhouette and it is impossible to say precisely what the pink touches
are meant to represent. Could it be that Jan van Eyck paid less attention to details
as small as this fortress in a very large work like the Ghent Altarpiece, which was
intended to be viewed from a greater distance? This argument can be dismissed on
the grounds that the altarpiece is full of small details – not least in the architecture
on the very horizon to which these mountains belong. Various elements seem to
indicate, therefore, that the ultramarine-based mountains and sky were added after
the landscape and buildings were fi nished. Stylistic comparison suggests that this
addition cannot be attributed to Jan.
We have seen in chapter 2 that a relatively major change was made at some point
in the groups in the foreground, where the kneeling and standing fi gures meet
161
4. ONE PAINTER OR SEVERAL?
4.13a 4.13b
162
4. ONE PAINTER OR SEVERAL?
4.15a 4.15b
Fig. 4.15a–b. Adoration of the Lamb, red chaperon
of a philosopher attributed to Jan van Eyck (a);
pink chaperon of a philosopher, non-Eyckian (b).
163
4. ONE PAINTER OR SEVERAL?
4.16a 4.16b
Fig. 4.16a–b.
Adoration of the
the pink chaperon, but the resulting grid is fl at and in no way follows the volume Lamb, green fringes
of the shoulder (cve-4). A similar change is found in the group to the right of the at the bottom of the
fountain, where the red robe of the third bishop has been enlarged to position it churchmen’s cloaks,
in front of rather than behind the kneeling apostle (cve-14). While the bishop non-Eyckian (a),
can be attributed to Jan, this enlargement does not seem painted by him. The attributed to Jan
green fringes at the bottom of the cloak (fi g. 4.16a) are represented by a number of van Eyck (b).
nondescript lines, unlike a similar detail lower down, in which the thickness and
texture of the fringed border are tangible (fi g. 4.16b).
Areas in which fi gures have been altered are also found in the wings, most notably
in the Knights of Christ. The knight with the grey beard wore a brocade robe, over
which a plain one was later painted (fi g. 2.66). This alteration can probably be
linked to the addition of a muzzle, meant to complete the head of a white horse in
the Just Judges (fi g. 4.17a–b). Comparison with the muzzles painted on the right in
the Knights of Christ clearly shows that this addition cannot have been Jan’s work.47
In order to paint the purple robe over the brocade, the painter fi rst narrowed
the edge of the beautiful green sleeve of the knight in armour. The ma-xrf maps
show that the vermilion base layer of this reworking (ma-xrf Hg-L), overlaps the
already painted green sleeve (ma-xrf Cu-K) (fi g. 4.18). Then, after the purple robe
had been glazed, the edge of the narrowed sleeve was reworked with more repetitive
and much less precise strokes than those placed elsewhere by Jan to indicate the
164
4. ONE PAINTER OR SEVERAL?
Non-Eyckian interventions seem to have been made for a variety of reasons. Some
of the retouches were apparently intended to ‘improve’ Jan’s work,51 such as those
along the boundary between the fi rst-stage fi gures painted by Hubert and the
second-stage meadow done by Jan. Others seem to have been placed in order to fi ll
empty spaces – such as the leaves added to many of the trees – or to create a certain
continuity between the different panels, as in the case of the horse’s muzzle and a
165
4. ONE PAINTER OR SEVERAL?
Fig 4.18a–h. Knights of Christ. In order to paint the purple robe over the brocade, the non-
Eyckian painter narrowed the edge of the green sleeve. The ma-xrf maps show that the
vermilion base layer of this reworking (Hg-L and red in the diagrams) overlaps the already
painted green sleeve (Cu-K and green lines in the diagrams). After the purple robe had been
completed (white arrow), the edge of the green sleeve was reworked (vis and diagram).
166
4. ONE PAINTER OR SEVERAL?
4.19a
4.19b 4.19c
Fig. 4.19a–c.
Comparison of birds’
heads: Adoration patch of grass on the crags on the left of the Knights of Christ, or the mountains to
of the Lamb, detail the right of the same panel.52 Some changes were also made to the organization of
of the dove (a); space, as in the perspective adjustment of the altar or the more marked separation
Virgin Annunciate between the kneeling and standing fi gures in the foreground of the central panel.
from the altarpiece An iconological or iconographical motive might also be sought in the addition
exterior, detail of
of gilded rays in the sky and around the head of the lamb.53 In one instance we
the dove (b); Virgin
of Canon Van der
encountered reworking in an area where a lacuna is embedded in the underlying
Paele, 1436, Bruges, paint layers (fi g. 2.63),54 and further research may eventually allow damage to be
Groeningemuseum, detected elsewhere too, as in the sky in the central panel. It is, indeed, still not
inv. 0000
0000.gro00161.i, known why the ultramarine layer was applied to the sky in the Adoration of the Lamb.
detail of the It is not possible to determine at this time whether these non-Eyckian interventions
parakeet (c). were the work of one person or several. We consider it unlikely that this reworking
was carried out by an assistant of Jan, given the implausibility of the master allowing
a subordinate to ‘improve’ his own work. It should be borne in mind that these
non-Eyckian retouches were covered by the repaints removed during the 20016–19
conservation campaign, which dated from the mid-sixteenth century (before 1557
at any rate).55
167
4. ONE PAINTER OR SEVERAL?
4.20
0a 4.20
0b
Fig. 4.200a–b.
Conclusion Adoration of the
Lamb, clear black
lines in the dove (b)
This stylistic study, in combination with the technical analysis, supports our similar to those
hypothesis that the lower register of the altarpiece interior as it appears today was found in the dove
the result of three successive stages: the fi rst by Hubert, the second by Jan, and the of the Annunciation
third post-dating the Van Eyck’s work. While we have paid particular attention to on the altarpiece
these non-Eyckian interventions, they are nevertheless quite limited: most of the exterior (a).
painted surface now visible in the lower register is the work of Jan van Eyck. Jan
did not retain the design of the spring, which he replaced with a majestic fountain.
He did, however, keep about fourteen faces around this central area, which can be
attributed to his brother Hubert.
168
4. ONE PAINTER OR SEVERAL?
Notes
1 See Introduction. 21 Martens 20019, pp. 127–28; Borchert
2 Dhanens 19800; Campbell 1998a, p. 174; 20019, pp. 148–500.
http://closertovaneyck.kikirpa.be/. 22 It is interesting to note that Hubert also
3 Including the Virgin of Chancellor Rolin, took account of the incidence of light
Paris, Musée du Louvre, inv. 1271; and from the right in the altarpiece’s original
the Annunciation, Washington, National location.
Gallery of Art, inv. 1937.1.39. 23 The reserves are visible in the xr and in
4 Campbell 20017; Martens 200200. the ma-xrf map.
5 Jan van Eyck, The Virgin at the Fountain, 24 Years of experience (visual observation,
1439, Antwerp, Royal Museum of Fine including stereomicroscopy and
Arts, inv. 411. X-radiography) have taught us that this
method is also found in other fi fteenth-
6 To compare the details, see http://
century Flemish masters, although as
closertovaneyck.kikirpa.be/.
far as we know it has not been described
7 Jan van Eyck, Arnolfini Portrait, 1434, in detail in the literature. Cennino
London, The National Gallery, inv. Cennini advised painters ‘to work on
ng186. clothes and buildings before faces’
8 Jan van Eyck, The Virgin of Canon Van der (Broecke 20015, p. 187). A similar sequence
Paele, 1436, Bruges, Groeningemuseum, of pictorial execution is found in certain
inv. 0000
0000.gro00161.i. unfi nished miniatures, such as folios 41v
9 See fi g. 2.22. and 55 in Guido delle Colonne, History of
the Destruction of Troy, painted around 1500
00
10 Campbell 20017. by François Colombe (Paris, Bibliothèque
11 The nose of this apostle and the top of his nationale de France, naf 249200).
forehead were later retouched, as we will 25 Jan van Eyck, St Barbara, 1437, Antwerp,
see in due course. Royal Museum of Fine Arts, inv. 410.
12 See http://closertovaneyck.kikirpa.be/. Van Hout 20012, pp. 34–37. When he
13 See chapter 2, sections ‘Meadow and died, Dirk Bouts left two works that
fountain’ and ‘The large fi gures in the displayed both fi nished and unfi nished
central panel’. areas; Stroo et al. 1999, pp. 56–104,
14 Especially the grass beneath the Madonna see pp. 76–78; Steyaert 200 005.
in the Virgin at the Fountain. See also 26 Hubert worked in Ghent and
‘Sequence of pictorial execution’ and died, according to his epitaph, on
note 24. 1 8 September 1 426. Jan was employed by
15 The apostles’ robes were reworked locally. John of Bavaria until the latter’s death in
The Hague on 5 January 1425, at which
16 Cross-section 10/81. point he immediately moved to Bruges.
17 Jan and Hubert had a sister, Margaretha He entered Philip the Good’s service on
(died before 1426), and a brother, 1 9 March 1425, but was allowed to accept
Lambert. Both were painters. None of other commissions. Sometime before
their work is known to us today. Unlike 2 August 1425, the duke ordered him
Hubert, they are not mentioned in the to move to Lille, where he remained for
quatrain. Paviot 200200, pp. 600, 79. three years. Paviot 200200, pp. 61–66.
18 Other authors have detected Hubert’s 27 See http://closertovaneyck.kikirpa.be/.
hand in the groups in the foreground 28 Depoorter 200200b, p. 52.
of the central panel, but without citing
specifi c areas and offering only general 29 Depoorter 200200a, p. 215.
arguments: Dvorák 19004; Panofsky 1953, 30 Jan van Eyck, Triptych of the
vol. 1, pp. 2005–32; for a summary, see Virgin and Child, 1437, Staatliche
Kemperdick 20014, pp. 19–22; Martens Kunstsammlungen Dresden, inv. 799.
20019, pp. 126–28. 31 See chapter 3, section ‘Ultramarine’ and
19 Cross-section 09/200. fi g. 3.10.
20 See chapter 2, sections ‘Meadow and 32 These characteristics were not dealt
fountain’ and ‘Soils in the wings’. with in our technical study of the
169
4. ONE PAINTER OR SEVERAL?
paintings of the altarpiece exterior (Postec, 44 The brush that applied the blue layer left
Steyaert 200200, pp. 200
00–216). We hope clearly visible striations. Similar, albeit
that the fi nal study, which will follow slightly less defi ned, brushstrokes can be
the treatment of the upper register of made out in the sky of the City View of the
the altarpiece interior, will provide an Ghent Altarpiece exterior and in the Virgin
opportunity to take stock of this issue. at the Fountain. The white, slightly
33 Panofsky 1953, p. 223. impastoed touches that defi ne the clouds
34 Van Asperen de Boer 1979, pp. 2005–13; and the way the moon is represented in
Périer-D’Ieteren 20017, pp. 1200–35. St Barbara are comparable with similar
details in the Knights of Christ. In
35 Van Asperen de Boer, Faries 19900, pp. 38, St Barbara, the birds were drawn before
43, 46; Van Asperen de Boer 1995, p. 81; the blue layer was applied over the drawing
Faries 1999, pp. 223–25; Gifford, Metzger, and then the birds were painted (http://
Delaney 20013, p. 133; Billinge 2000000, pp. closertovaneyck.kikirpa.be/: compare
94–95; Périer-D’Ieteren 20017, pp. 122, the macrophotograph with the infrared
123, 127. macrophotograph to the left of the tower).
36 See chapter 2. These features were not 45 Comblen-Sonkes, Lorentz 1995, ii, p. 58.
as visible in the images (the Virgin of 46 The milky pupils of this philosopher
Chancellor Rolin) analysed by Van Asperen appear to suggest some kind of visual
de Boer 1979, pp. 2008–10. In the new impairment.
irrs (see http://closertovaneyck.kikirpa.
be/), these small, transverse lines cannot be 47 The head of the horse on the right has
found in the Virgin of Chancellor Rolin. been painted with subtle nuances of colour
and every anatomical detail has been
37 The drawing is less visible in the irrs of precisely observed – the veins, the fold of
the clothes of the other fi gures, possibly the nostril, the wrinkles in the lips, and
due to a higher concentration of pigments even the hair on the muzzle. These are all
that absorb infrared or the larger number details and nuances that are entirely absent
of paint layers. in the horse on the left.
38 See http://closertovaneyck.kikirpa.be/. 48 Panofsky 1953, pp. 218–19; Coremans
St John the Baptist and the Virgin and Child, 1953, p. 108; Van Asperen de Boer 1979,
Paris, Musée du Louvre, inv. rf 1938–22. pp. 195–96 and 200 00–2001.
39 Comblen-Sonkes, Lorentz 1995, pp. 5 49 See http://closertovaneyck.kikirpa.be/.
and 7. Hand, Metzger, Spronk 200 006–007,
pp. 78–79. 50 The dove is highly likely to have been
present in the Ghent Altarpiece before 1458,
40 See chapter 2, section ‘Trees’. when the polyptych was staged as a tableau
41 Campbell 20017; Postec, Steyaert 200200, vivant. Dubois, Sanyova, Vanwijnsberghe
p. 234. 20017, p. 73.
42 See note 24. 51 Regarding the restoration of works in the
43 It might appear in the Virgin at the fi fteenth and sixteenth centuries,
Fountain that the blue upper layer of the see Campbell 1998b, p. 21.
sky was laid down after the fi gures were 52 See chapter 2, sections ‘Trees’, ‘The large
painted, as the brushstrokes follow the fi gures in the wings’ and ‘Rocks in the
contours of the fi gures (the wings of the wings’.
angels, for instance). However, the sky was 53 See also chapter 1, note 23.
painted fi rst, around the contours of the
fi gures indicated in the underdrawing. We 54 See chapter 2, sections ‘The sky and the
also fi nd that the painted sky runs around horizon’ and ‘ The large fi gures in the
the forms of the drawing in Jan van Eyck’s central panel’ and note 124.
St Barbara. See further: Van Asperen de 55 Dubois 20018a; Depuydt et al. 200200,
Boer 1979, p. 195, note 71. pp. 124–35; see also chapter 1.
170
Conclusion
The recent conservation campaign carried out on the paintings of the lower register
of the altarpiece interior and the removal of several layers of degraded varnish
and the sixteenth-century overpaints revealed the incredible richness of the Van
Eycks’ work, while also allowing it to be studied. The fi ndings of that research add
signifi cantly to the knowledge previously gained regarding the Ghent Altarpiece.
Nevertheless, a great deal remains to be explored. We very much hope that this data
– which will no doubt be supplemented further by the forthcoming conservation of
the upper register of the altarpiece interior – will offer pointers for future research.
The restoration campaign carried out in 19500–51 by Paul Coremans and Albert
Philippot was itself accompanied by a technical and scientifi c study, which provided
important information from interdisciplinary work. At that point, however, it
was not possible to determine the extent of the overpaints, due to the presence of
many layers of degraded varnish and to analysis techniques that were less advanced
than those of today. Consequently, earlier research featured zones that were largely
covered by sixteenth-century overpaints, the presence of which was not suspected
at the time.
In the six decades that separate Coremans and Philippot’s campaign from the one
commenced in 20012 and not yet completed, our understanding of historic pictorial
techniques – fi fteenth-century Flemish painting in this instance – has advanced
considerably. Scientifi c progress continues to expand our knowledge of the materials
making up the pictorial layers, pigments and binders, but also of how they were
manufactured and how they age. Not to mention the presence of additives, the very
existence of which was not even suspected until recently. The interactions between
all these materials, lastly, are likewise understood better today, allowing a more
refi ned or more nuanced interpretation of the analysis results.
All the data collected in the course of the years required for the restoration of
the paintings of the lower register of the altarpiece interior have been carefully
examined both during and after treatment, enabling us to present a number of
hypotheses. Some had been outlined before, but never expanded on in such
detail – particularly as regards the share of the Van Eyck brothers in the work.
(facing page) Even though many questions remain, the degree to which the altarpiece was
Hermits, heads unfi nished on Hubert’s death in 1426 has begun to emerge. It is not always easy
attributed to Jan to distinguish the pictorial layers attributable to each stage of intervention. What
van Eyck (see also is clear, however, is that several layers have been added to an already complete
fi g. 4.6). stratigraphy in a number of locations. In other words, a fi rst stage consisting of
173
CONCLUSION
several layers and completed with a surface glaze was then covered with another
stratigraphy comprising multiple layers and ending once more with a glaze. Two
distinct stages of development can be found in the meadow, for instance. Study of
the scientifi c imagery – particularly ma-xrf scans and infrared macrophotographs
– as well as examination under the stereomicroscope of painted layers discernible
in losses beneath the currently visible composition, also allowed this fi rst stage to
be visualized elsewhere. Differences have been detected in, among other places, the
horizon line – where other mountains and architecture had already been painted
and possibly even completed – and the soil and rocks in the shutters. Determining
the authorship of these interventions remains a challenge. Nevertheless, stylistic
comparison suggests that the second stage can be attributed to Jan and the fi rst
stage, which remains visible in certain places on the surface, is different and might
be Hubert’s.
It seems quite possible, therefore, that the fi rst layers of painting of the entire
lower register of the altarpiece interior, which correspond with the fi rst stage, had
largely been completed by the time of Hubert’s death, implying that all the work
in the preparatory layers was fi nished in 1426, including the underdrawing. Jan
van Eyck not only fi nished the paintings, he also made numerous adjustments.
Some of these are formal in character: he repainted some of the buildings on the
horizon and added others; he reworked many of the garments and altered the
position of some of the large fi gures and certain horses, while adjusting the soil.
He also reworked the landscape, which he transformed by painting a new meadow
in the central panel with a varied and abundant fl ora. Other changes were more
signifi cant, notably in the central axis of the Adoration of the Lamb panel. In the fi rst
stage, there was a simple spring in the earth below the altar, which Jan covered
in the second stage with a majestic fountain. It would seem that the dove we see
now was not planned in the sky in the fi rst stage – only rays of light that were not
executed, possibly because that could only be done when the work was otherwise
complete. Jan abandoned the rays and, although we have no proof, it might be to
him that we owe the addition of the dove.
There are rays in the sky, however, that were covered by the now-removed
sixteenth-century overpaint. They are far more numerous than the initially
planned rays, increasing from around 28 to 129. We think that – like the current
dove, the mediocrity of which cannot be attributed to Jan – the rays, too, were
added by a painter other than the Van Eycks. This represents a third stage in
the development of the paintings of the interior lower register as it appears today
following conservation. The three stages observed here were detected because they
are superimposed, but they also occur in juxtaposition. Laboratory analyses of
samples do not allow the execution of the dove and the rays to be dated precisely,
but they do show that they are contemporary and that they differ from the materials
used elsewhere. The fi nal blue layer of sky in the central panel, lastly, on which
the gilded rays are laid, and the many touches scattered over the paintings of the
lower register of the altarpiece interior seem to us to have been added during this
third stage. These adjustments alter the perspective – especially in the altar of the
Adoration of the Lamb – they ensure continuity between the different shutters and
dispel a certain horror vacui. Certain changes were likewise made to the organization
174
CONCLUSION
of space, as in the more marked separation between the four groups of fi gures in
the foreground of the central panel. All the indications are that these additions also
post-date the second stage, over which they have been placed. Laboratory analyses
have so far been unable to demonstrate that they were done after the Van Eycks, but
stylistic comparison suggests that they can hardly be attributed to Jan. They were
superimposed in areas completed by him. It seems unlikely that they were done
by assistants, since it is barely credible that they would have altered the master’s
fi nished work. It can be assumed that most of these non-Eyckian reworkings date
from the second half of the fi fteenth century or the beginning of the sixteenth
at the latest. These additions have been retained during the recent conservation
campaign.
The extra year of in-depth study enabled us to collate the team’s observations and
also to make fresh discoveries. This prospecting work deserves to be continued. The
question of the lacunae observed in the lower strata of the paintings, for example, has
not been completely resolved, and their extent has yet to be evaluated. Combining
the fi ndings of both the abundant scientifi c imagery and the meticulous observation
of the paintings is, after all, a long-term process. We hope to have demonstrated,
however, that it can result in signifi cant discoveries.
175
Photography before
and after Treatment
All high-resolution images, before, during and after treatment, as well as scientifi c
imagery, are available in open access on http://closertovaneyck.kikirpa.be. They
have been made by the kik-irpa team, Sophie De Potter, Katrien Van Acker,
Catherine Fondaire, Stéphane Bazzo and Hervé Pigeolet, under the direction of
Christina Currie.
(facing page)
The photographers
of kik-irpa make
an ongoing visual
record of the work
as a whole and its
details in normal,
raking, ultraviolet
and infrared light.
177
178
179
180
181
188
189
190
191
Bibliography
193
BIBLIOGRAPHY
Specifi c Signatures by Raman and Infrared Spectroscopy Taniguchi, Laurence De Viguerie, Manfred Burghammer,
for Their Characterization in Medieval Artworks’, in Philippe Walter, Camille Rivard, Murielle Salomé, Koen
Heritage, 2, 20019, pp. 1614–29. Janssens and Jean Susini, ‘Lead soaps in paintings: Friends
Caen et al. 200006 or foes?’, in Studies in Conservation, 62 (1), 20016, pp. 2–23.
Joost Caen, Olivier Schalm and Koen Janssens, ‘15th Currie 200
009
century stained glass windows in the former county of Christina Currie, ‘Genesis of a Pre-Eyckian Masterpiece:
Flanders; a historical and chemical study related to recent Melchior Broederlam’s Painted Wings for the Crucifi xion
conservation campaigns’, in Koen Janssens et al. (eds), Altarpiece’, in Stroo 200009, pp. 23–86.
Annales du 17e Congrès de l’Association Internationale pour Currie et al. 20017
l’Histoire du Verre, Antwerp, 200 006, pp. 459–66. Christina Currie, Bart Fransen, Valentine Henderiks,
Campbell 1998a Cyriel Stroo and Dominique Vanwijnsberghe (eds), Van
Lorne Campbell, The Fifteenth-Century Netherlandish Schools Eyck Studies. Papers Presented at the Eighteenth Symposium
(National Gallery Catalogues), New Haven/London, 1998. for the Study of Underdrawing and Technology in Painting,
Campbell 1998b Brussels, 19–21 September 2012 (Underdrawing and Technology
Lorne Campbell, ‘The conservation of Netherlandish in Painting. Symposium, 18), Paris/Louvain/Bristol, CT,
Paintings in the Fifteenth and Sixteenth Centuries’, in 20017.
Christine Sitwell and Sarah Staniforth (eds), Studies in De Mey et al. 20012
the History of Painting Restoration, 1998, pp. 15–26. Marc De Mey, Maximiliaan P.J. Martens and Cyriel Stroo
Campbell 20017 (eds), Vision and Material. Interaction between Art and Science
Lorne Campbell, ‘The Speed of Illusion’, in Currie et al. in Jan van Eyck’s Time, KVAB Press (Koninklijke Vlaamse
20017, pp. 256–61. Academie van België voor Wetenschappen en Kunsten),
Campbell, Foister, Roy 1997a Brussels, 20012, pp. 235–53.
Lorne Campbell, Susan Foister and Ashok Roy (eds), Deneffe, Vanwijnsberghe 20019
‘Methods and materials of Northern European Painting Dominique Deneffe and Dominique Vanwijnsberghe (eds),
in the National Gallery, 1400 00–15500’, in National Gallery ‘A Man of Vision’. Paul Coremans and the Preservation of
Technical Bulletin, 18, 1997, pp. 6–55. Cultural Heritage Worldwide. Proceedings of the International
Campbell, Foister, Roy 1997b Symposium Paul Coremans Held in Brussels, 15–17 June 2015
Lorne Campbell, Susan Foister and Ashok Roy (eds), ‘The (Scientia Artis, 15), Brussels, 20019, pp. 1200–39.
materials and technique of fi ve paintings by Rogier van der Deneffe, Peters, Fremout 200 009
Weyden and his Workshop’, in National Gallery Technical Dominique Deneffe, Famke Peters and Wim Fremout,
Bulletin, 18, 1997, pp. 68–86. Pre-Eyckian Panel Painting in the Low Countries, 1. Catalogue
Cennini (ed. Thomson) 19600 (Contributions to Fifteenth-Century Painting in the Southern
The Craftsman’s Handbook. ‘II libro dell’Arte’ by Cennino Netherlands and the Principality of Liège, 9), Brussels,
d’Andrea Cennini, D.V. Thompson (ed.), New York, 19600. 200
009.
Claes et al. 20019 Depoorter 200200a
Marie-Christine Claes, Hélène Dubois and Jana Sanyova, Matthias Depoorter, ‘Jan Van Eyck’s Discovery of Nature’,
‘Le quotidien d’un idéal: L’Agneau mystique, catalyseur de in Martens et al. 200200b, pp. 2005–35.
l’interdisciplinarité’, in Deneffe, Vanwijnsberghe 20019, Depoorter 200200b
pp. 1400–57. Matthias Depoorter, ‘Birds on the Ghent altarpiece’,
Clarke 20 011 in Van den Bremt, Van Crombrugge 200200, pp. 45–52.
Mark Clarke, Medieval painters materials and techniques. Depuydt-Elbaum et al. 200200
The Montpellier Liber diversarum arcium, London, 20011. Livia Depuydt-Elbaum, Françoise Rosier, Bart Devolder
Comblen-Sonkes, Lorentz 1995 and Nathalie Laquière, Conservation and Restoration
Micheline Comblen-Sonkes and Philippe Lorentz, Le Musée Treatment, The Painted Surface, in Fransen, Stroo 200200,
du Louvre Paris (Corpus de la peinture des anciens Pays-Bas pp. 111–66.
méridionaux et de la Principauté de Liège au quinzième siècle, De Schryver, Marijnissen 1953
1 7), Brussels, 1 995. Antoine De Schryver and Roger Henri Marijnissen,
Coremans 1953 Histoire matérielle, in Coremans 1953, pp. 21–68.
Paul Coremans, L’Agneau mystique au Laboratoire. Examen Dhanens 19800
et traitement (Contributions à l’étude des Primitifs flamands, 2), Elisabeth Dhanens, Hubert en Jan van Eyck, Antwerp, 19800.
Antwerp, 1953. Dietz et al. 20016
Coremans, Thissen 1953 Stephanie Dietz, Christoph Krekel, Andrea Obermeier and
Paul Coremans and Jean Thissen, ‘Matériaux originaux Heike Stege, ‘Ground glass in Holbein the Elder’s work’,
et technique picturale eyckienne’, in Coremans 1953, in Arie Wallert (ed.), Painting techniques: history, materials
pp. 69–76. and studio practice, 5th International Symposium, Rijksmuseum,
Coremans, Loose, Thissen 1953 Amsterdam, 1820 September 2013, Rijksmuseum,
Paul Coremans, Louis Loose and Jean Thissen, ‘Quelques Amsterdam, 20016, pp. 43–47.
problèmes particuliers’, in Coremans 1953, pp. 98–122. Dubois 20017
Cotte et al. 20016 Hélène Dubois, ‘Michiel Coxcie’s Copy as a Formal
Marine Cotte, Emilie Checroun, Wout De Nolf, Yoko Reference of the Material Condition of the Ghent
194
BIBLIOGRAPHY
Altarpiece in 1557’, in Stephan Kemperdick, Johannes Wetenschappen, Letteren en Schoone Kunsten van België. Klasse
Rößler and Joris Corin Heyder (eds), Der Genter der Schoone Kunsten, 7, 4), Antwerp, 1945, pp. 1–89.
Altar. The Ghent Altarpiece. Reproduktionen, Deutungen, Dvorák 19004
Forschungskontroversen. Reproductions, Interpretations, Scholarly Max Dvorák, ‘Das Rätsel der Kunst der Brüder Van Eyck’,
Debates, Berlin/Petersberg, 20017, pp. 78–93. in Jahrbuch der kunsthistorischen Sammlungen des allerhöchsten
Dubois 20018a Kaiserhauses, 24, 5, 19004, pp. 161–318.
Hélène Dubois, ‘The Art of Conservation XV. The Eastaugh et al. 200 004
Conservation History of the Ghent Altarpiece’, in Nicholas Eastaugh, Valentine Walsh, Tracey Chaplin and
The Burlington Magazine, 1600, 1386, September 20018, Ruth Siddall, The Pigment Compendium: A Dictionary of
pp. 754–65. Historical Pigments, Amsterdam, 200 004.
Dubois 20018b Faries 1997
Hélène Dubois, ‘The Adoration of the Lamb’ Molly Faries, ‘Jan van Scorel’s Clerical Patronage’, in
http://balat.kikirpa.be/doc/pdf/Saisies_147.pdf, Bolletino d’arte, Suppl. vol. 100
00, 1997, pp. 107–16.
‘The Virgin, The Deity and John the Baptist enthroned’ Faries 1999
http://balat.kikirpa.be/doc/pdf/Saisies_0059.pdf. Molly Faries, ‘The Underdrawing in Jan van Eyck’s
in Inventaris van de schilderijen en beeldhouwwerken die ten Dresden Triptych’, in Hélène Verougstraete and Roger
tijde van de Franse Revolutie in België werden in beslag genomen Van Schoute (eds), La peinture dans les Pays-Bas au 16e siècle.
en naar Frankrijk overgebracht (1794–1795)’, kik-irpa, Pratiques d’atelier. Infrarouges et autres méthodes d’investigation.
Brussels. http://balat.kikirpa.be/tools/saisies/indexnl.html. Symposium 11–13 septembre 1997 (Le dessin sous-jacent et
Dubois 200200a la technologie dans la peinture, 12), Louvain-la-Neuve, 1999,
Hélène Dubois, ‘Transformations in the Sixteenth and pp. 221–29.
Seventeenth Centuries’, in Fransen, Stroo 200200, pp. 11–46. Faries 20011
Dubois 200200b Molly Faries, ‘Jan van Scorel’s Drawing and Painting
Hélène Dubois, ‘When, by Whom and Why? Decisive Technique’, in Liesbeth M. Helmus and Dorien
Material and Optical Alterations of the Ghent Altarpiece’, Tamis (eds), Utrecht Painting, 1363–1600 00. The Collection
in Martens et al. 200200, pp. 237–57. of the Centraal Museum, Utrecht, 20011, pp. 22–42.
Dubois, Deneffe 20019 Fondaire, Vanwijnsberghe 200 008
Hélène Dubois and Dominique Deneffe, ‘Construction Catherine Fondaire and Dominique Vanwijnsberghe,
of a Vision: Coremans’ Approach to the Study of Flemish ‘Van der Veken et Van Eyck aux rayons X’, in Dominique
Primitives before the Restoration of the Adoration of Vanwijnsberghe (ed.) in collaboration with Catherine
the Mystic Lamb (The Ghent Altarpiece)’, in Deneffe, Bourguignon and Jacques Debergh, Autour de la Madeleine
Vanwijnsberghe 20019, pp. 1200–39. Renders. Un aspect de l’histoire des collections, de la restauration
Dubois et al. 200
009 et de la contrefaçon en Belgique dans la première moitié du
Anne Dubois, Roel Slachmuylders, Géraldine Patigny and XXe siècle (Scientia Artis, 4), Brussels, 200 008, pp. 283–86.
Famke Peters, The Flemish Primitives V. Anonymous Masters Fraiture 20017
(Catalogue of Early Netherlandish Painting in the Royal Pascale Fraiture, ‘Results of Three Campaigns of
Museums of Fine Arts of Belgium), Brussels, 200 009. Dendrochronological Analysis on the Ghent Altarpiece
Dubois, Sanyova, Vanwijnsberghe 20017 (1986–20013)’, in Currie et al. 20017, pp. 76–95.
Hélène Dubois, Jana Sanyova and Dominique Fransen 20017
Vanwijnsberghe, ‘“Revenons à notre mouton”. Paul Bart Fransen, ‘Van Eyck in Valencia’, in Currie et al. 20017,
Coremans, Erwin Panofsky, Martin Davies and the Mystic pp. 469–78.
Lamb’, in Currie et al. 20017, pp. 2700–79 and pp. 66–75. Fransen, Stroo 200200
Dumolyn, Buylaert 200200 Bart Fransen and Cyriel Stroo (eds), The Ghent Altarpiece –
Jan Dumolyn and Frederik Buylaert, ‘Van Eyck’s World’, Research and Conservation of the Exterior (Contributions to the
in Martens et al. 200200b, pp. 85–121. Study of the Flemish Primitives, 14), Brussels, 200200.
Dunkerton 20011 Froyen, Dubois 200200
Jill Dunkerton, ‘Leonardo in Verrochio’s Workshop: Kathleen Froyen and Hélène Dubois (eds), Conservation
Re-examining the Technical Evidence’, in National Gallery and restoration treatment of the Adoration of the Mystic Lamb
Technical Bulletin, 32, 20011, pp. 4–31. polyptych (The Ghent Altarpiece); Phase 2: the lower register
Dunkerton, Billinge 200 005 of the open altarpiece (20016–20019), Brussels, kik-irpa,
Jill Dunkerton and Rachel Billinge, Beyond the naked eye. April 200200. http://data.closertovaneyck.be/ec2/data/
Details from the National Gallery, London, 200 005. GhentAltarpiece_treatment_Phase2.pdf.
Dunkerton, Morrison, Roy 20017 Gifford 1999
Jill Dunkerton, Rachel Morrison and Ashok Roy, Melanie E. Gifford, ‘Van Eyck’s Washington
‘Pigments, Media and Varnish Layers on the Portrait of Annunciation: Technical Evidence for Iconographic
Margaret van Eyck’, in Currie et al. 20017, pp. 2700–800. Development’, in The Art Bulletin, 81, 1999, 1,
Duverger 1945 pp. 108–16.
Jozef Duverger, ‘Het grafschrift van Hubrecht Van Eyck Gifford 2000000
en het quatrain van het Gentsche Lam Gods-retabel’ Melanie E. Gifford, ‘Assessing the Evolution of Van Eyck’s
(Verhandelingen van de Koninklijke Vlaamsche Academie voor Iconography through Technical Study of the Washington
195
BIBLIOGRAPHY
Annunciation, I.’, in Delphine Cool et al., Investigating Jan The Ghent Altarpiece by the Brothers Van Eyck. History and
van Eyck, Turnhout, 2000 000, pp. 59–66. Appraisal, Berlin, 20014, pp. 8–69.
Gifford et al. 20017 Kemperdick, Lammertse 20012
Melanie E. Gifford, John K. Delaney, Suzanne Quillen Stephan Kemperdick and Friso Lammertse, The Road to
Lomax, Rachel Morrison and Marika Spring, Van Eyck, Rotterdam, 20012.
‘New Findings on the Painting Medium of the Ketels, Glatigny, Augustyniak 200200
Washington Annunciation’, in Currie et al. 20017, Jochen Ketels, Jean-Albert Glatigny and Anne-Sophie
pp. 281–900. Augustyniak, ‘Frames and Support: Technique and
Gifford, Metzger, Delaney 20013 Structural Treatment’, in Fransen, Stroo 200200,
Melanie E. Gifford, Catherine A. Metzger and John pp. 47–76.
K. Delaney, ‘Jan van Eyck’s Washington Annunciation: Keune, Boon 200 005
painting materials and techniques’, in Daphne Barbour Katrien Keune and Jaap J. Boon, ‘Analytical imaging
and Melanie E. Gifford (eds), Facture: conservation, science, studies clarifying the process of the darkening of vermilion
art history, Renaissance Masterworks, 1, 20013, pp. 128–53. in paintings’, in Analytical chemistry, 77, 15, 200
005,
Glanville 1995 pp. 4742–500.
Helen Glanville, ‘Varnish, Grounds, Viewing Distance, Kirby 200
008
and Lighting: Some Notes on 17th-century Italian Jo Kirby, ‘Some Aspects of Medieval and Renaissance Lake
Painting Technique’, in Arie Wallert, Erma Hermens and Pigment Technology’, in Dyes in History and Archaeology,
Marja Peek (eds), Historical Painting Techniques, Materials, 21, 200
008, pp. 89–108.
and Studio Practice. Preprints of a Symposium, University Kirby 20012
of Leiden, the Netherlands, 26–29 June 1995, The Getty Jo Kirby, ‘Aspects of Oil-Painting in Northern
Conservation Institute, Los Angeles, 1995, pp. 12–19. Europe and Jan van Eyck’, in De Mey et al. 20012,
Glatigny et al. 20010 pp. 255–78.
Jean-Albert Glatigny et al., ‘Report of the research on Kirby et al. 20014
the structural condition of the panels and the frames of Jo Kirby, Maarten van Bommel, André Verhecken and
the Ghent Altarpiece, 20010’. http://legacy.closertovaneyck. Marika Spring, Natural Colorants for Dyeing and Lake
be/#home/sub=documents. Pigments: Practical Recipes and Their Historical Sources,
Gonzalez et al. 20017 London, 20014.
Victor Gonzalez, Didier Gourier, Thomas Calligaro, Kirby, Saudners, Spring 200 006
Kathleen Toussaint, Gilles Wallez and Michel Menu, Jo Kirby, David Saudners and Marika Spring, ‘Proscribed
‘Revealing the Origin and History of Lead-White Pigments in Northern European Renaissance Paintings
Pigments by their Photoluminescence Properties’, and the Case of Paris Red’, in Studies in Conservation, 51,
in Analytical Chemistry, 89, 20017, pp. 29009–18. Suppl. 2, 200006, pp. 236–43.
Hand, Metzger, Spronk 200 006–007 Kirby, Spring, Higgitt 200 005
John Oliver Hand, Catherine A. Metzger and Ron Spronk, Jo Kirby, Marika Spring and Catherine Higgitt, ‘The
Prayers and Portraits: Unfolding the Netherlandish Diptych, Technology of Red Lake Pigment Manufacture: Study
Antwerp, 200 006–007. of the Dyestuff Substrate’, in National Gallery Technical
Hedley 1993 Bulletin, 26, 200
005, pp. 71–87.
Gerry Hedley, ‘Long Lost Relations and New Found Klaassen et al. 20021
Relativities. Issues in the Cleaning of Paintings’, in Lizet Klaassen, Marie Postec, Geert Van der Snickt and
Caroline Villers (ed.), Measured Opinions, Collected Papers on Marika Spring, ‘Materials and Painting Technique of
the Conservation of Paintings, London, 1993, pp. 172–78. Memling’s Nájera Panels’, in Lizet Klaassen and Dieter
Janssens et al. 1998 Lampens (eds), Harmony in Bright Colors. Memling’s God
Koen Janssens, Ine Deraedt, Olivier Schalm and the Father with Singing and Music-Making Angels Restored,
Johan Veeckman, ‘Composition of 15–17th Century Turnhout, 20021, pp. 86–123.
Archaeological Glass Vessels Excavated in Antwerp, Kockaert 1995
Belgium’, in Microchimica Acta, 15, 1998, pp. 253–67. Léopold Kockaert, ‘Problems concerning the Brussels
Jones, Augustyniak, Dubois 200200 Lamentation by Petrus Christus’, in Maryan
Susan Frances Jones, Anne-Sophie Augustyniak and W. Ainsworth (ed.), Petrus Christus in Renaissance Bruges,
Hélène Dubois, ‘The Authenticity of the Quatrain and an interdisciplinary approach, 1995, pp. 189–92.
the other Frame Inscriptions’, in Fransen, Stroo 200200, Kühnen, Hern 20016
pp. 273–3007. Renate Kühnen and Christoph Hern, ‘Protein fi bres as
Keller 1973 intermediate layer on medieval shields, panel paintings
Renate Keller, ‘Leinöl als Malmittel: and altarpieces’, in Zeitschrift für Kunsttechnologie und
Rekonstruktionsversuche nach Rezepten aus dem 13. Konservierung, 3, 20016, pp. 36–46.
bis 19. Jahrhundert’, in Maltechnik Restauro, 79, 1973, Lutzenberger, Stege, Tilenschi 20010
pp. 74–105. Karin Lutzenberger, Heike Stege und Cornelia Tilenschi,
Kemperdick 20014 ‘A note on new fi ndings of glass and silica in oil paintings
Stephan Kemperdick, ‘The History of the Ghent from the 15th to the 17th century’, in Journal of Cultural
Altarpiece’, in Stephan Kemperdick and Johannes Rößler, Heritage, 11, 20010, pp. 365–72.
196
BIBLIOGRAPHY
197
BIBLIOGRAPHY
198
BIBLIOGRAPHY
199
Paintings’, in Susan Foister, Sue Jones and Delphine Cool
(eds), Investigating Jan Van Eyck, Turnhout, 2000
000,
pp. 107–17.
Von Baum et al. 20014
Katja Von Baum, Patrick Dietemann, Ulrike Fischer, Peter
Klein, Roland Krischel, Theresa Neuhoff, Caroline Von
Saint-George, Iris Schaefer, Martin Schawe, Heike Stege
and Jeanine Walcher, Let the Material Talk: Technology of
Late-medieval Cologne Panel Painting, Cologne/Munich,
20014.
Project Participants
Participants of the 20016–20019 conservation/restoration Ingrid Depoorter (ace), Philippe Depotter (ba), Bart
treatment of the Ghent Altarpiece (phase 2). The affi liation Devolder (kik/irpa), Hélène Dubois (kik/irpa, ugent), Hans
given corresponds to the participant’s function within the Feys (ace), Kathleen Froyen (kik/irpa), Lieven Gerard (msk),
project. Maximiliaan Martens (ugent), Pascale De Meyer (msk), Jan
Mortier (aoe), Martine Pieteraerens (pov), Jo Rombouts
(pov), Johan De Smet (msk), Wesley De Smet (msk), Cyriel
The Project Stroo (kik/irpa), Johan Vandenberghe (kksb), Cathérine
Verleysen (msk), Peter De Wilde (tv, kksb), Catherine
Commissioning authority De Zegher (msk)
Churchwardens of St Bavo’s Cathedral, Ghent (kksb)
Site monitoring committee
Project planning Griet Blanckaert (ua), Till-Holger Borchert (gm, vkc),
Bressers Architects bvba, Ghent (ba), with Christina Christina Ceulemans (kik/irpa), Hilde De Clercq (kik/irpa),
Ceulemans (kik/irpa), Hilde De Clercq (kik/irpa), Ludo Ludo Collin (kksb), Ingrid Depoorter (ace, coordination),
Collin (kksb), Maximiliaan Martens (ugent) and Ron Spronk Bart Devolder (kik/irpa), Hélène Dubois (kik/irpa, ugent),
(qu, run) Hans Feys (ace), Koen Janssens (ua), Evelien Oomen (aoe)
Working group art history
Administrative support
Till-Holger Borchert (gm, vkc), Bart Fransen (kik/irpa,
Johan De Ridder (kksb), Johan Vandenberghe (kksb) coordination), Maximiliaan Martens (ugent), Ron Spronk
(qu, run), Cyriel Stroo (kik/irpa)
Advisory committee
Liesbeth De Belie (kmskb/mrbab), Christiaan Beyaert Working group communication
(kksb), Maaike Blancke (Bba), Till-Holger Borchert (gm, Nele Bogaert (pov, coordination), Tim Bottelberghe (tov),
vkc), Véronique Bücken (kmskb/mrbab), Gonda Callaert Catherine Bourguignon (kik/irpa), Bart Devolder (kik/
(aoe), Christina Ceulemans (kik/irpa), Annemie Charlier irpa), Valerie Evers (rwo), Kathleen Froyen (kik/irpa),
(pov), Hilde De Clercq (kik/irpa), Ludo Collin (kksb), Dirk Simon Laevers (kik/irpa), Stephanie Lenoir (ugent), Bart
Cuvelier (g), Ingrid Depoorter (ke), Philippe Depotter (ba, Ooghe (msk), Evelien Oomen (aoe), Martine Pieteraerens
chair), Bart Devolder (kik/irpa), Hélène Dubois (kik/irpa, (pov), Johan De Ridder (kksb), Jabez Sinnesael (pcc), Birgit
ugent), Hans Feys (ace), Bart Fransen (kik/irpa), Kathleen Vandamme (rwo), Ann Vervaeke (ugent), Ben De Vriendt
Froyen (kik/irpa), Jean-Albert Glatigny (asajag), Anne van (kksb, ttv), Céline De Waegenaere (tg), Leen Van Wezemael
Grevenstein (tr, uva), Natalie Honraet (ba), Margriet Van (cjsm), Erwin van de Wiele (tg)
Houtte (kksb), Beatrice Janssens (kksb), Andrea De Kegel
(mov), Jochen Ketels (kik/irpa), Christian Lardinoit (kksb), Working group education
Thomas Leysen (tr), Jacques Lust (belspo), Madeleine
Manderyck (aoe), Maximiliaan Martens (ugent), Catheline Bart Devolder (kik/irpa), Kathleen Froyen (kik/irpa),
Metdepenninghen (aoe), Pascale De Meyer (msk), Halewijn Lies Ledure (msk), Anneke Lippens (pov), Els Otte (pov,
Missiaen (aoe), Timothy Naessens (laf), Anne-Cathérine coordination), Jabez Sinnesael (pcc), Amand Vilyn (kksb)
Olbrechts (pov), Evelien Oomen (aoe), Philip Oosterlinck
(kksb), Martine Pieteraerens (pov), Jo Rombouts (pov), Working group relocation
Peter Schmidt (kul), Johan De Smet (msk), Wesley De Smet Till-Holger Borchert (mb, vkc), Ingrid Depoorter (ace),
(msk), Ron Spronk (qu, run), Cyriel Stroo (kik/irpa), Peter Philippe Depotter (ba), Hans Feys (ace), Anne van
Vandenabeele (ugent), Johan Vandenberghe (kksb), Cathérine Grevenstein (tr, uva), Jan Mortier (aoe), Anne-Cathérine
Verleysen (msk), Hélène Verougstraete (formerly ucl), Ben Olbrechts (faro), Mira van Olmen (aoe), Philip Oosterlinck
De Vriendt (kksb, tv), Jørgen Wadum (smk), Catherine De (kksb), Jan Van der Stock (tr, kul), Lieve Watteeuw (tr,
Zegher (msk) kul)
201
PROJECT PARTICIPANTS
Bart Fransen (kik/irpa), Kathleen Froyen (kik/irpa), Margriet Higgitt (ngl), Koen Janssens (ua, coordination), Štěpánka
Van Houtte (kksb), Simon Laevers (kik/irpa), Maximiliaan Kučková (vŠcht), Stijn Legrand (ua), Francisco Mederos-
Martens (ugent), Timothy Naessens (af), Edwin De Roock Henry (kik/irpa), Claude Poleunis (ucl), Jana Sanyova (kik/
(kik/irpa), Ron Spronk (qu, run, coordination), Frederik irpa, coordination), Geert Van der Snickt (ua), Marika Spring
Temmermans (ud, imec-vub) (ngl), Frederik Vanmeert (ua)
202
PROJECT PARTICIPANTS
203
Photographic Acknowledgements
All illustrations of the Ghent Altarpiece are made by Antwerp, Royal Museum of Fine Arts: fi g. 4.3b.
kik-irpa, Brussels. The copyright of these images belongs Musea Brugge – Groeningemuseum, Lukasweb.be –
to: Sint-Baafskathedraal © Lukasweb.be – Art in Flanders Art in Flanders vzw, photo kik-irpa: fi g. 4.2c, 4.4b, 4.19c.
vzw, photo kik-irpa. Stratigraphic cross-sections of the Brussels, kik-irpa, Polychrome Artefact Laboratory: fi g. 3.1,
Ghent Altarpiece belong to kik-irpa’s Polychrome Artefact 3.2a, 3.2b, 3.2c, 3.2d, 3.3a, 3.3b, 3.3c, 3.3d, 3.4 b, 3.4c,
Laboratory. Diagrams belong to the authors. Exceptions and 3.5a, 3.5b, 3.5c, 3.5d, 3.6, 3.10 b, 3.10c, 3.11b, 3.11c,
other illustration credits are listed below. Every effort has 3.11d, 3.12d, 3.12e, 3.12f, 3.14b, 3.15b, 3.15c, 3.15d,
been made to contact copyright-holders of illustrations. 3.16b, 3.17, 3.18f, 3.18g, 3.22a, 3.22b, 3.22c, 3.22d,
Any copyright-holder whom we have been unable to reach 3.22e, 3.22f, 3.22g, 3.22h, 3.22i, 3.22j, 3.22k, 3.22l.
or to whom inaccurate acknowledgement has been made is Dresden, Staatliche Kunstsammlungen: fi g. 4.8b.
invited to contact the publisher. London, © The National Gallery, London: fi g. 4.1b, 4.14a.
New York, Metropolitan Museum of Art: fi g. 1.26.
Paris, Bibliothèque nationale de France: fi g. 2.15.
Antwerp University: fi g. 1.8b, 2.14a, 2.31b, 2.35c, 2.39c, Paris, Musée du Louvre: fi g. 4.2a, 4.4c, 4.12b, 4.13b.
2.39d, 2.39e, 2.41a, 2.42b, 2.47a, 2.52b, 2.53b, 2.53c, © RMN-Grand Palais (domaine de Chantilly) / Michel
2.59b, 2.63b, 2.66b, 2.66d, 2.68, 2.72b, 3.4d, 3.4e, 3.4f, Urtado: fi g. 2.16.
3.4g, 3.7b, 3.7c, 3.7d, 3.7e, 3.7f, 3.9b, 3.9c, 3.9d, 3.9e,
3.9f, 3.12b, 3.12c, 3.13b, 3.13c, 3.13d, 3.16c, 3.16d, 3.16e,
3.16f, 3.16g, 3.16h, 3.18b, 3.18c, 3.18d, 3.18e, 3.19a,
3.19b, 3.19c, 3.19d, 3.19f, 3.200b, 3.200c, 3.21b, 3.21c,
4.18c, 4.18g.
204
Abbreviations
205
The Ghent Altarpiece by Hubert and Jan van Eyck has always attracted both admiration
and curiosity. Following the treatment of the closed polyptych, we turned to the paintings
of the lower register of the open altarpiece, which includes the famous Adoration of the
Lamb. A major research and restoration campaign by the Royal Institute for Cultural
Heritage (kik-irpa, Brussels) was carried out between 2016 and 2020.
The findings reflect those from the treatment of the closed altarpiece: large areas of the
panels were covered with overpaints, the vast majority dating from the 16th century. Their
removal allows the Van Eycks brothers’ paintings to be truly appreciated again.
The three years that the conservators spent working on the panels enabled them to
examine the paintings down to the smallest detail. Interdisciplinary collaboration with
laboratory scientists, art historians and experts in scientific imagery led to a series of
discoveries – some anticipated in light of earlier research, others completely unexpected.
The authenticity of the quatrain stating that Hubert had begun the work, that his
brother Jan completed it and that it was presented in 1432, has been firmly established
recently. New hypotheses are proposed here regarding the division of labour between the
two Van Eyck brothers. The contribution of Hubert, who died in 1426, can finally be seen
more clearly and turns out to be far from negligible. But the removal of overpaints has
revealed non-Eyckian interventions too...
978-2-930054-41-4
ISBN 978-2-39025-441-4